R: Difference between revisions

From 太極
Jump to navigation Jump to search
(834 intermediate revisions by the same user not shown)
Line 1: Line 1:
= Install and upgrade R =
= Install and upgrade R =
[[Install_R|Here]]
[[Install_R|Here]]
== New release ==
* R 4.3.0
** [https://www.jumpingrivers.com/blog/whats-new-r43/ What's new in R 4.3.0?]
** Extracting from a pipe. The underscore _ can be used to refer to the final value from a pipeline <code style="display:inline-block;">mtcars |> lm(mpg ~ disp, data = _) |> _$coef</code>. Previously we need to use [https://stackoverflow.com/a/56038303 this way] or [https://stackoverflow.com/a/60873298 this way]. If we want to apply some (anonymous) function to each element of a list, use '''map(), map_dbl()''' from the [https://purrr.tidyverse.org/ purrr].
* R 4.2.0
** Calling if() or while() with a condition of length greater than one gives an error rather than a warning.
** [https://twitter.com/henrikbengtsson/status/1501306369319735300 use underscore (_) as a placeholder on the right-hand side (RHS) of a forward pipe]. For example, '''mtcars |> subset(cyl == 4) |> lm(mpg ~ disp, data = _) '''
** [https://developer.r-project.org/Blog/public/2022/04/08/enhancements-to-html-documentation/ Enhancements to HTML Documentation]
** [https://www.jumpingrivers.com/blog/new-features-r420/ New features in R 4.2.0]
* R 4.1.0
** [https://developer.r-project.org/blosxom.cgi/R-devel/2021/01/13#n2021-01-13 pipe and shorthand for creating a function]
** [https://www.jumpingrivers.com/blog/new-features-r410-pipe-anonymous-functions/ New features in R 4.1.0] '''anonymous functions''' (lambda function)
* R 4.0.0
** [https://blog.revolutionanalytics.com/2020/04/r-400-is-released.html R 4.0.0 now available, and a look back at R's history]
** [https://www.infoworld.com/article/3540989/major-r-language-update-brings-big-changes.html R 4.0.0 brings numerous and significant changes to syntax, strings, reference counting, grid units, and more], [https://www.infoworld.com/article/3541368/how-to-run-r-40-in-docker-and-3-cool-new-r-40-features.html R 4.0: 3 new features]
**# factor is not default in data frame for character vector
**# palette() function has a new default set of colours, and [[R#New_palette_in_R_4.0.0|palette.colors() & palette.pals()]] are new
**# r"(YourString)" for ''raw'' character constants. See ?Quotes
* R 3.6.0
** [https://blog.revolutionanalytics.com/2019/05/whats-new-in-r-360.html What's new in R 3.6.0]
*** Changes to random number generation
*** More functions now support vectors with more than 2 billion elements
* R 3.5.0
** [https://community.rstudio.com/t/error-listing-packages-error-in-readrds-pfile-cannot-read-workspace-version-3-written-by-r-3-6-0/40570/2 The default serialization format for R changed in May 2018, such that new default format (version 3) for workspaces saved can no longer be read by versions of R older than 3.5]


= Online Editor =
= Online Editor =
Line 22: Line 47:


= Web Applications =
= Web Applications =
[[R_web|R web applications]]


See also CRAN Task View: [http://cran.r-project.org/web/views/WebTechnologies.html Web Technologies and Services]
= Creating local repository for CRAN and Bioconductor =
[[R_repository|R repository]]


== TexLive ==
= Parallel Computing =
TexLive can be installed by 2 ways
See [[R_parallel|R parallel]].
* '''sudo apt install texlive''' It includes '''tlmgr''' utility for package manager.
* [http://tug.org/texlive/ Official website]


=== texlive-latex-extra ===
= Cloud Computing =
https://packages.debian.org/sid/texlive-latex-extra


For example, framed and titling packages are included.
== Install R on Amazon EC2 ==
http://randyzwitch.com/r-amazon-ec2/


=== tlmgr - TeX Live package manager ===
== Bioconductor on Amazon EC2 ==
https://www.tug.org/texlive/tlmgr.html
http://www.bioconductor.org/help/bioconductor-cloud-ami/


== [https://yihui.name/tinytex/ TinyTex] ==
= Big Data Analysis =
https://github.com/yihui/tinytex
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
* [https://rstudio-pubs-static.s3.amazonaws.com/72295_692737b667614d369bd87cb0f51c9a4b.html Handling large data sets in R]
* [https://www.oreilly.com/library/view/big-data-analytics/9781786466457/#toc-start Big Data Analytics with R] by Simon Walkowiak
* [https://pbdr.org/publications.html pbdR]
** https://en.wikipedia.org/wiki/Programming_with_Big_Data_in_R
** [https://olcf.ornl.gov/wp-content/uploads/2016/01/pbdr.pdf Programming with Big Data in R - pbdR] George Ostrouchov and Mike Matheson Oak Ridge National Laboratory


== Rmarkdown: create HTML5 web, slides and more ==
== bigmemory, biganalytics, bigtabulate ==
[[Rmarkdown]]


== [http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol HTTP protocol] ==
== ff, ffbase ==
* tapply does not work. [https://stackoverflow.com/questions/16470677/using-tapply-ave-functions-for-ff-vectors-in-r Using tapply, ave functions for ff vectors in R]
* [http://www.bnosac.be/index.php/blog/12-popularity-bigdata-large-data-packages-in-r-and-ffbase-user-presentation Popularity bigdata / large data packages in R and ffbase useR presentation]
* [http://www.bnosac.be/images/bnosac/blog/user2013_presentation_ffbase.pdf ffbase: statistical functions for large datasets] in useR 2013
* [https://www.rdocumentation.org/packages/ffbase/versions/0.12.7/topics/ffbase-package ffbase] package


* http://en.wikipedia.org/wiki/File:Http_request_telnet_ubuntu.png
== biglm ==
* [http://en.wikipedia.org/wiki/Query_string Query string]
* How to capture http header? Use '''curl -i en.wikipedia.org'''.
* [http://trac.webkit.org/wiki/WebInspector Web Inspector]. Build-in in Chrome. Right click on any page and choose 'Inspect Element'.
* [http://en.wikipedia.org/wiki/Web_server Web server]
* [http://www.paulgriffiths.net/program/c/webserv.php Simple TCP/IP web server]
* [http://jmarshall.com/easy/http/ HTTP Made Really Easy]
* [http://www.manning.com/hethmon/ Illustrated Guide to HTTP]
* [http://www.ibm.com/developerworks/systems/library/es-nweb/ nweb: a tiny, safe Web server with 200 lines]
* [http://sourceforge.net/projects/tinyhttpd/ Tiny HTTPd]


An HTTP server is conceptually simple:
== data.table ==
See [[Tidyverse#data.table|data.table]].


# Open port 80 for listening
== disk.frame ==
# When contact is made, gather a little information (get mainly - you can ignore the rest for now)
[https://www.brodrigues.co/blog/2019-10-05-parallel_maxlik/ Split-apply-combine for Maximum Likelihood Estimation of a linear model]
# Translate the request into a file request
# Open the file and spit it back at the client


It gets more difficult depending on how much of HTTP you want to support - POST is a little more complicated, scripts, handling multiple requests, etc.
== Apache arrow ==
* https://arrow.apache.org/docs/r/
* [https://www.infoworld.com/article/3637038/the-best-open-source-software-of-2021.html#slide17 The best open source software of 2021]


=== Example in R ===
= Reproducible Research =
<syntaxhighlight lang='r'>
* http://cran.r-project.org/web/views/ReproducibleResearch.html
> co <- socketConnection(port=8080, server=TRUE, blocking=TRUE)
* [[Reproducible|Reproducible]]
> # Now open a web browser and type http://localhost:8080/index.html
> readLines(co,1)
[1] "GET /index.html HTTP/1.1"
> readLines(co,1)
[1] "Host: localhost:8080"
> readLines(co,1)
[1] "User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0"
> readLines(co,1)
[1] "Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
> readLines(co,1)
[1] "Accept-Language: en-US,en;q=0.5"
> readLines(co,1)
[1] "Accept-Encoding: gzip, deflate"
> readLines(co,1)
[1] "Connection: keep-alive"
> readLines(co,1)
[1] ""
</syntaxhighlight>


=== Example in C ([http://blog.abhijeetr.com/2010/04/very-simple-http-server-writen-in-c.html Very simple http server written in C], 187 lines) ===
== Reproducible Environments ==
https://rviews.rstudio.com/2019/04/22/reproducible-environments/


Create a simple hello world html page and save it as <[http://en.wikipedia.org/wiki/List_of_Hello_world_program_examples#H index.html]> in the current directory (/home/brb/Downloads/)
== checkpoint package ==
* https://cran.r-project.org/web/packages/checkpoint/index.html
* [https://timogrossenbacher.ch/2017/07/a-truly-reproducible-r-workflow/ A (truly) reproducible R workflow]


Launch the server program (assume we have done ''gcc http_server.c -o http_server'')
== Some lessons in R coding ==
<pre>
# don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use [[Rcpp|Rcpp]] package and R's random number generator instead.
$ ./http_server -p 50002
# don't use [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/list.files list.files()] directly. The result is platform dependent even different Linux OS. An extra [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/sort sorting] helps!
Server started at port no. 50002 with root directory as /home/brb/Downloads
</pre>


Secondly open a browser and type http://localhost:50002/index.html. The server will respond
= Useful R packages =
<pre>
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
GET /index.html HTTP/1.1
* [https://github.com/qinwf/awesome-R awesome-R]
Host: localhost:50002
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/index.html
== Rcpp ==
GET /favicon.ico HTTP/1.1
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].
Host: localhost:50002
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/favicon.ico
== RInside : embed R in C++ code ==
GET /favicon.ico HTTP/1.1
* http://dirk.eddelbuettel.com/code/rinside.html
Host: localhost:50003
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/favicon.ico
=== Ubuntu ===
</pre>
With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on [http://www.youtube.com/watch?v=UQ8yKQcPTg0 Youtube]. I have tested this '''qtdensity''' example successfully using Qt 4.8.5.
The browser will show the page from <index.html> in server.
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
# Install [[Qt|Qt]]. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
# Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
# Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the ''interactive'' program qtdensity appears on your desktop.


The only bad thing is the code does not close the port. For example, if I have use Ctrl+C to close the program and try to re-launch with the same port, it will complain '''socket() or bind(): Address already in use'''.
[[:File:qtdensity.png]]


With RInside + [http://www.webtoolkit.eu/wt Wt web toolkit] installed, we can also create a web application. To demonstrate the example in ''examples/wt'' directory, we can do
<pre>
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
make
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
</pre>
Then we can go to the browser's address bar and type ''http://localhost:8080'' to see how it works (a screenshot is in [http://dirk.eddelbuettel.com/blog/2011/11/30/ here]).


=== Another Example in C (55 lines) ===
=== Windows 7 ===
http://mwaidyanatha.blogspot.com/2011/05/writing-simple-web-server-in-c.html
To make RInside works on Windows OS, try the following
 
# Make sure R is installed under '''C:\''' instead of '''C:\Program Files''' if we don't want to get an error like ''g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory''.
The response is embedded in the C code.  
# Install RTools
 
# Instal RInside package from source (the binary version will give an [http://stackoverflow.com/questions/13137770/fatal-error-unable-to-open-the-base-package error ])
If we test the server program by opening a browser and type "http://localhost:15000/", the server received the follwing 7 lines
# Create a DOS batch file containing necessary paths in PATH environment variable
<pre>
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
cmd
</pre>
In the Windows command prompt, run
<pre>
<pre>
GET / HTTP/1.1
cd C:\R\R-3.0.1\library\RInside\examples\standard
Host: localhost:15000
make -f Makefile.win
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive
</pre>
</pre>
 
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
If we include a non-executable file's name in the url, we will be able to download that file. Try "http://localhost:15000/client.c".
 
If we use telnet program to test, wee need to type anything we want
<pre>
<pre>
$ telnet localhost 15000
rinside_sample0
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
ThisCanBeAnything        <=== This is what I typed in the client and it is also shown on server
HTTP/1.1 200 OK          <=== From here is what I got from server
Content-length: 37Content-Type: text/html
 
HTML_DATA_HERE_AS_YOU_MENTIONED_ABOVE <=== The html tags are not passed from server, interesting!
Connection closed by foreign host.
$
</pre>
</pre>


See also more examples under [[C#Socket_Programming_Examples_using_C.2FC.2B.2B.2FQt|C page]].
As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See  some discussions in
 
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
=== Others ===
* http://www.mail-archive.com/rcpp-[email protected]-forge.r-project.org/msg04377.html
* http://rosettacode.org/wiki/Hello_world/ (Different languages)
So the Qt and Wt web tool applications on Windows may or may not be possible.
* http://kperisetla.blogspot.com/2012/07/simple-http-web-server-in-c.html (Windows web server)
* http://css.dzone.com/articles/web-server-c (handling HTTP GET request, handling content types(txt, html, jpg, zip. rar, pdf, php etc.), sending proper HTTP error codes, serving the files from a web root, change in web root in a config file, zero copy optimization using sendfile method and php file handling.)
* https://github.com/gtungatkar/Simple-HTTP-server
* https://github.com/davidmoreno/onion


== shiny ==
== GUI ==
See [[Shiny|Shiny]].
=== Qt and R ===
* http://cran.r-project.org/web/packages/qtbase/index.html [https://stat.ethz.ch/pipermail/r-devel/2015-July/071495.html QtDesigner is such a tool, and its output is compatible with the qtbase R package]
* http://qtinterfaces.r-forge.r-project.org


== [https://www.rplumber.io/ plumber]: Turning your R code into a RESTful Web API ==
== tkrplot ==
* https://github.com/trestletech/plumber
On Ubuntu, we need to install tk packages, such as by
* https://www.rstudio.com/resources/videos/plumber-turning-your-r-code-into-an-api/
<pre>
* [https://blog.rstudio.com/2018/10/23/rstudio-1-2-preview-plumber-integration/ RStudio 1.2 Preview: Plumber Integration]
sudo apt-get install tk-dev
* [https://medium.com/@skyetetra/using-docker-to-deploy-an-r-plumber-api-863ccf91516d Using docker to deploy an R plumber API]
</pre>


== Docker ==
== reticulate - Interface to 'Python' ==
* There are two major Docker images. They include gcc, gfortran, .... So it can be used to install Rcpp package for example.
[[Python#R_and_Python:_reticulate_package|Python -> reticulate]]
** [https://hub.docker.com/_/r-base/?tab=tags Official] which supports version tags. The official Docker image's dockerfile still points to rocker/r-base.
** [https://hub.docker.com/r/rocker/r-base rocker project] which only has the latest tag
* [https://blog.ouseful.info/2016/05/03/using-docker-as-a-personal-productvity-tool-running-command-line-apps/ Using Docker as a Personal Productivity Tool – Running Command Line Apps Bundled in Docker Containers]
* [https://peerj.com/preprints/3181.pdf#page=8 Dockerized RStudio server] from Duke University. 110 containers were set up on a cloud server (4 cores, 28GB RAM, 400GB disk). Each container has its own port number. Each student is mapped to a single container. https://github.com/mccahill/docker-rstudio
* [http://sas-and-r.blogspot.com/2016/12/rstudio-in-cloud-with-amazon-lightsail.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+SASandR+%28SAS+and+R%29 RStudio in the cloud with Amazon Lightsail and docker]
* Mark McCahill (RStudio + Docker)
** http://sites.duke.edu/researchcomputing/files/2014/09/mccahill-DockerDays.pdf
** https://github.com/mccahill/docker-rstudio
** https://hub.docker.com/r/mccahill/rstudio/~/dockerfile/
* [https://github.com/Bioconductor-notebooks/BiocImageBuilder BiocImageBuilder]
** [https://github.com/Bioconductor-notebooks/Identification-of-Differentially-Expressed-Genes-for-Ectopic-Pregnancy/blob/master/CaseStudy1_EctopicPregnancy.ipynb Reproducible Bioconductor Workflow w/ browser-based interactive notebooks+Container].
** [http://biorxiv.org/content/early/2017/06/01/144816 Paper]
** Original [http://www.rna-seqblog.com/reproducible-bioconductor-workflows-using-browser-based-interactive-notebooks-and-containers/ post].
* [https://www.opencpu.org/posts/opencpu-with-docker/ Why Use Docker with R? A DevOps Perspective]
* [https://www.statworx.com/de/blog/running-your-r-script-in-docker/ Running your R script in Docker]. Goal: containerizing an R script to eventually execute it automatically each time the container is started, without any user interaction. An enhanced version of the instruction is at [https://github.com/arraytools/RinDocker this page].


== [http://cran.r-project.org/web/packages/httpuv/index.html httpuv] ==
== Hadoop (eg ~100 terabytes) ==
http and WebSocket library.
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]


See also the [https://cran.r-project.org/web/packages/servr/index.html servr] package which can start an HTTP server in R to serve static files, or dynamic documents that can be converted to HTML files (e.g., R Markdown) under a given directory.
* RHadoop
* Hive
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].
* http://www.techspritz.com/category/tutorials/hadoopmapredcue/ Single node or multinode cluster setup using Ubuntu with VirtualBox (Excellent)
* [http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/ Running Hadoop on Ubuntu Linux (Single-Node Cluster)]
* Ubuntu 12.04 http://www.youtube.com/watch?v=WN2tJk_oL6E and [https://www.dropbox.com/s/05aurcp42asuktp/Chiu%20Hadoop%20Pig%20Install%20Instructions.docx instruction]
* Linux Mint http://blog.hackedexistence.com/installing-hadoop-single-node-on-linux-mint
* http://www.r-bloggers.com/search/hadoop


== [http://rapache.net/ RApache] ==
=== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ===
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
* Ubuntu 12.04 - [http://crishantha.com/wp/?p=1414 Crishantha.com], [http://nikhilshah123sh.blogspot.com/2014/03/setting-up-rhadoop-in-ubuntu-1204.html nikhilshah123sh.blogspot.com].[http://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/ Bighadoop.wordpress] contains an example.
* RapReduce in R by [https://github.com/RevolutionAnalytics/rmr2/blob/master/docs/tutorial.md RevolutionAnalytics] with a few examples.
* https://twitter.com/hashtag/rhadoop
* [http://bigd8ta.com/step-by-step-guide-to-setting-up-an-r-hadoop-system/ Bigd8ta.com] based on Ubuntu 14.04.


== [http://cran.r-project.org/web/packages/gWidgetsWWW/index.html gWidgetsWWW] ==
=== Snowdoop: an alternative to MapReduce algorithm ===
* http://matloff.wordpress.com/2014/11/26/how-about-a-snowdoop-package/
* http://matloff.wordpress.com/2014/12/26/snowdooppartools-update/comment-page-1/#comment-665


* http://www.jstatsoft.org/v49/i10/paper
== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
* [https://github.com/jverzani/gWidgetsWWW2 gWidgetsWWW2] gWidgetsWWW based on Rook
On Ubuntu, we need to install libxml2-dev before we can install XML package.
* [http://www.r-statistics.com/2012/11/comparing-shiny-with-gwidgetswww2-rapache/ Compare shiny with gWidgetsWWW2.rapache]
<pre>
sudo apt-get update
sudo apt-get install libxml2-dev
</pre>


== [http://cran.r-project.org/web/packages/Rook/index.html Rook] ==
On CentOS,
<pre>
yum -y install libxml2 libxml2-devel
</pre>


Since R 2.13, the internal web server was exposed.
=== XML ===
 
* http://giventhedata.blogspot.com/2012/06/r-and-web-for-beginners-part-ii-xml-in.html. It gave an example of extracting the XML-values from each XML-tag for all nodes and save them in a data frame using '''xmlSApply()'''.
[https://docs.google.com/present/view?id=0AUTe_sntp1JtZGdnbjVicTlfMzFuZDQ5dmJxNw Tutorual from useR2012] and [https://github.com/rstats/RookTutorial Jeffrey Horner]
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
* https://www.tutorialspoint.com/r/r_xml_files.htm
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
* [http://www.stat.berkeley.edu/~statcur/Workshop2/Presentations/XML.pdf Extracting data from XML] PubMed and Zillow are used to illustrate. xmlTreeParse(),  xmlRoot(),  xmlName() and xmlSApply().
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
{{Pre}}
library(XML)


Here is another [http://www.rinfinance.com/agenda/2011/JeffHorner.pdf one] from http://www.rinfinance.com.
# Read and parse HTML file
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)


Rook is also supported by [rApache too. See http://rapache.net/manual.html.
# Extract all the paragraphs (HTML tag is p, starting at
# the root of the document). Unlist flattens the list to
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


Google group. https://groups.google.com/forum/?fromgroups#!forum/rrook
# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)


Advantage
# Join all the elements of the character vector into a single
* the web applications are created on desktop, whether it is Windows, Mac or Linux.
# character string, separated by spaces
* No Apache is needed.
doc.text = paste(doc.text, collapse = ' ')
* create [http://jeffreyhorner.tumblr.com/post/4723187316/introducing-rook multiple applications] at the same time. This complements the limit of rApache.
 
----
 
4 lines of code [http://jeffreybreen.wordpress.com/2011/04/25/4-lines-of-r-to-get-you-started-using-the-rook-web-server-interface/ example].
 
<pre>
library(Rook)
s <- Rhttpd$new()
s$start(quiet=TRUE)
s$print()
s$browse(1)  # OR s$browse("RookTest")
</pre>
</pre>
Notice that after s$browse() command, the cursor will return to R because the command just a shortcut to open the web page http://127.0.0.1:10215/custom/RookTest.


[[File:Rook.png|100px]]
This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.
[[File:Rook2.png|100px]]
{{Pre}}
[[File:Rookapprnorm.png|100px]]
> library(RCurl) # getURL()
> library(XML)  # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # <span class="css-truncate-target">2.5.3a</span>
> plain.text
[1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"   
[6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"


We can add Rook '''application''' to the server; see ?Rhttpd.
> # try picard
<pre>
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
s$add(
> doc = htmlParse(xData)
    app=system.file('exampleApps/helloworld.R',package='Rook'),name='hello'
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
)
[1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
s$add(
[10] "2.6.0"
    app=system.file('exampleApps/helloworldref.R',package='Rook'),name='helloref'
</pre>
)
This method can be used to monitor new tags/releases from some projects like [https://github.com/Ultimaker/Cura/releases Cura], BWA, Picard, [https://github.com/alexdobin/STAR/releases STAR]. But for some projects like [https://github.com/ncbi/sra-tools sratools] the '''class''' attribute in the '''span''' element ("css-truncate-target") can be different (such as "tag-name").
s$add(
    app=system.file('exampleApps/summary.R',package='Rook'),name='summary'
)


s$print()
=== xmlview ===
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/


#Server started on 127.0.0.1:10221
== RCurl ==
#[1] RookTest http://127.0.0.1:10221/custom/RookTest
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
#[2] helloref http://127.0.0.1:10221/custom/helloref
{{Pre}}
#[3] summary  http://127.0.0.1:10221/custom/summary
# Test on Ubuntu 14.04
#[4] hello    http://127.0.0.1:10221/custom/hello
sudo apt-get install libxml2-dev
 
sudo apt-get install libcurl4-openssl-dev
#  Stops the server but doesn't uninstall the app
## Not run:
s$stop()
 
## End(Not run)
s$remove(all=TRUE)
rm(s)
</pre>
</pre>
For example, the interface and the source code of ''summary'' app are given below


[[File:Rookappsummary.png|100px]]
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


<nowiki>
No google ID is required
app <- function(env) {
    req <- Rook::Request$new(env)
    res <- Rook::Response$new()
    res$write('Choose a CSV file:\n')
    res$write('<form method="POST" enctype="multipart/form-data">\n')
    res$write('<input type="file" name="data">\n')
    res$write('<input type="submit" name="Upload">\n</form>\n<br>')


    if (!is.null(req$POST())){
Seems not work
data <- req$POST()[['data']]
<pre>
res$write("<h3>Summary of Data</h3>");
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
res$write("<pre>")
  arguments imply differing number of rows: 2, 0
res$write(paste(capture.output(summary(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
</pre>
res$write("</pre>")
res$write("<h3>First few lines (head())</h3>");
res$write("<pre>")
res$write(paste(capture.output(head(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
res$write("</pre>")
    }
    res$finish()
}
</nowiki>


More example:
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
* http://lamages.blogspot.com/2012/08/rook-rocks-example-with-googlevis.html
'''devtools''' package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the [[#remotes|remotes]] package which was suggested by the [https://cran.r-project.org/web/packages/BiocManager/index.html BiocManager] package.
* [http://www.road2stat.com/cn/r/rook.html Self-organizing map]
{{Pre}}
* Deploy Rook apps with rApache. [http://jeffreyhorner.tumblr.com/post/27861973339/deploy-rook-apps-with-rapache-part-i First one] and [http://jeffreyhorner.tumblr.com/post/33814488298/deploy-rook-apps-part-ii two].
# Ubuntu 14.04
* [https://rud.is/b/2016/07/05/a-simple-prediction-web-service-using-the-new-firery-package/ A Simple Prediction Web Service Using the New fiery Package]
sudo apt-get install libcurl4-openssl-dev


== [https://code.google.com/p/sumo/ sumo] ==
# Ubuntu 16.04, 18.04
Sumo is a fully-functional web application template that exposes an authenticated user's R session within java server pages. See the paper http://journal.r-project.org/archive/2012-1/RJournal_2012-1_Bergsma+Smith.pdf.
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev


== [http://www.stat.ucla.edu/~jeroen/stockplot Stockplot] ==
# Ubuntu 20.04
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev
</pre>


== [http://www.rforge.net/FastRWeb/ FastRWeb] ==
[https://github.com/wch/movies/issues/3 Lazy-load database XXX is corrupt. internal error -3]. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.
http://cran.r-project.org/web/packages/FastRWeb/index.html


== WebDriver ==
NB. According to the output of '''apt-cache show r-cran-devtools''', the binary package is very old though '''apt-cache show r-base''' and [https://cran.r-project.org/bin/linux/ubuntu/#supported-packages supported packages] like ''survival'' shows the latest version.
'WebDriver' Client for 'PhantomJS'  


https://github.com/rstudio/webdriver
=== [https://github.com/hadley/httr httr] ===
httr imports curl, jsonlite, mime, openssl and R6 packages.


== [http://sysbio.mrc-bsu.cam.ac.uk/Rwui/tutorial/Instructions.html Rwui] ==
When I tried to install httr package, I got an error and some message:
 
<pre>
== [http://cran.r-project.org/web/packages/CGIwithR/index.html CGHWithR] and [http://cran.r-project.org/web/packages/WebDevelopR/ WebDevelopR] ==
Configuration failed because openssl was not found. Try installing:
CGHwithR is still working with old version of R although it is removed from CRAN. Its successor is WebDevelopR. Its The vignette (year 2013) provides a review of several available methods.
* deb: libssl-dev (Debian, Ubuntu, etc)
 
* rpm: openssl-devel (Fedora, CentOS, RHEL)
== [http://www.rstudio.com/ide/docs/advanced/manipulate manipulate] from RStudio ==
* csw: libssl_dev (Solaris)
This is not a web application. But the '''manipulate''' package can be used to create interactive plot within R(Studio) environment easily. Its source is available at [https://github.com/rstudio/rstudio/tree/master/src/cpp/r/R/packages/manipulate here].
* brew: openssl (Mac OSX)
 
If openssl is already installed, check that 'pkg-config' is in your
Mathematica also has manipulate function for plotting; see [http://reference.wolfram.com/mathematica/tutorial/IntroductionToManipulate.html here].
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
 
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
== [https://github.com/att/rcloud RCloud] ==
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
RCloud is an environment for collaboratively creating and sharing data analysis scripts. RCloud lets you mix analysis code in R, HTML5, Markdown, Python, and others. Much like Sage, iPython notebooks and Mathematica, RCloud provides a notebook interface that lets you easily record a session and annotate it with text, equations, and supporting images.
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’
</pre>
It turns out after I run '''sudo apt-get install libssl-dev''' in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!


See also the [http://user2014.stat.ucla.edu/abstracts/talks/193_Harner.pdf Talk] in UseR 2014.
Real example: see [http://stackoverflow.com/questions/27371372/httr-retrieving-data-with-post this post]. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).


== [https://github.com/cloudyr cloudyr] and [https://github.com/socialcopsdev/flyio flyio] - Input Output Files in R from Cloud or Local ==
Since httr package was used in many other packages, take a look at how others use it. For example, [https://github.com/ropensci/aRxiv aRxiv] package.
https://blog.socialcops.com/inside-sc/announcements/flyio-r-package-interact-data-cloud/ Announcing flyio, an R Package to Interact with Data in the Cloud]


== Dropbox access ==
[https://www.statsandr.com/blog/a-package-to-download-free-springer-books-during-covid-19-quarantine/ A package to download free Springer books during Covid-19 quarantine], [https://www.radmuzom.com/2020/05/03/an-update-to-an-adventure-in-downloading-books/ An update to "An adventure in downloading books"] (rvest package)
[https://cran.r-project.org/web/packages/rdrop2/index.html rdrop2] package


== Web page scraping ==
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
http://www.slideshare.net/schamber/web-data-from-r#btnNext
curl is independent of RCurl package.


=== [https://cran.r-project.org/web/packages/xml2/ xml2] package ===
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
rvest package depends on xml2.
* https://www.opencpu.org/posts/curl-release-0-8/


=== [https://cran.r-project.org/web/packages/purrr/index.html purrr]: Functional Programming Tools ===
{{Pre}}
* https://purrr.tidyverse.org/
library(curl)
* [http://colinfay.me/purrr-cookbook/ purrr cookbook]
h <- new_handle()
* Functional programming (cf Object-Oriented Programming)
handle_setform(h,
** [http://www.youtube.com/watch?v=vLmaZxegahk Functional programming for beginners]
  name="aaa", email="bbb"
** [https://www.makeuseof.com/tag/functional-programming-languages/ 5 Functional Programming Languages You Should Know]
)
* [http://data.library.virginia.edu/getting-started-with-the-purrr-package-in-r/ Getting started with the purrr package in R], especially the [https://www.rdocumentation.org/packages/purrr/versions/0.2.5/topics/map map()] function.
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
* [http://staff.math.su.se/hoehle/blog/2019/01/04/mathgenius.html Purr yourself into a math genius]
rawToChar(req$content)
* [https://martinctc.github.io/blog/vignette-write-and-read-multiple-excel-files-with-purrr/ Write & Read Multiple Excel files with purrr]
</pre>


=== [https://cran.r-project.org/web/packages/rvest/index.html rvest] ===
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
[http://blog.rstudio.org/2014/11/24/rvest-easy-web-scraping-with-r/ Easy web scraping with R]
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment


On Ubuntu, we need to install two packages first!
== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
<syntaxhighlight lang='bash'>
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).
sudo apt-get install libcurl4-openssl-dev # OR libcurl4-gnutls-dev


sudo apt-get install libxml2-dev
Example:
</syntaxhighlight>
{{Pre}}
# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')
</pre>


* https://github.com/hadley/rvest
== DirichletMultinomial ==
* [http://datascienceplus.com/visualizing-obesity-across-united-states-by-using-data-from-wikipedia/ Visualizing obesity across United States by using data from Wikipedia]
On Ubuntu, we do
* [https://stat4701.github.io/edav/2015/04/02/rvest_tutorial/ rvest tutorial: scraping the web using R]
<pre>
* https://renkun.me/pipeR-tutorial/Examples/rvest.html
sudo apt-get install libgsl0-dev
* http://zevross.com/blog/2015/05/19/scrape-website-data-with-the-new-r-package-rvest/
</pre>
* [https://datascienceplus.com/google-scholar-scraping-with-rvest/ Google scholar scraping with rvest package]


=== Animate ===
== Create GUI ==
* [https://guyabel.com/post/football-kits/ Animating Changes in Football Kits using R]: rvest, tidyverse, xml2, purrr & magick
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
* [https://guyabel.com/post/animated-directional-chord-diagrams/ Animated Directional Chord Diagrams] tweenr & magick
* [http://smarterpoland.pl/index.php/2019/01/x-mas-trees-with-gganimate-ggplot-plotly-and-friends/ x-mas tRees with gganimate, ggplot, plotly and friends]
* [https://www.listendata.com/2019/05/create-animation-in-r-learn-with.html Create animation in R]: learn by examples (gganimate)


=== [https://cran.r-project.org/web/packages/V8/index.html V8]: Embedded JavaScript Engine for R ===
== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
[https://rud.is/b/2017/07/25/r%E2%81%B6-general-attys-distributions/ R⁶ — General (Attys) Distributions]: V8, rvest, ggbeeswarm, hrbrthemes and tidyverse packages are used.
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]


=== [http://cran.r-project.org/web/packages/pubmed.mineR/index.html pubmed.mineR] ===
== json ==
Text mining of PubMed Abstracts (http://www.ncbi.nlm.nih.gov/pubmed). The algorithms are designed for two formats (text and XML) from PubMed.
[[R_web#json|R web -> json]]


[https://github.com/jtleek/swfdr R code for scraping the P-values from pubmed, calculating the Science-wise False Discovery Rate, et al] (Jeff Leek)
== Map ==
=== [https://rstudio.github.io/leaflet/ leaflet] ===
* rstudio.github.io/leaflet/#installation-and-use
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


== These R packages import sports, weather, stock data and more ==
=== choroplethr ===
* https://www.computerworld.com/article/3109890/data-analytics/these-r-packages-import-sports-weather-stock-data-and-more.html
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
* https://github.com/ALShum/rwunderground
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
* [https://tophcito.blogspot.com/2015/11/accessing-apis-from-r-and-little-r.html Accessing APIs from R (and a little R programming)]. A personal key is required. 500 times per day for a free account.
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/
* https://cran.r-project.org/web/packages/rnoaa/index.html. A personal API key (token) is required. 10,000 requests per day
* <strike>http://ram-n.github.io/weatherData/ </strike>  (not working)


== Diving Into Dynamic Website Content with splashr ==
=== ggplot2 ===
https://rud.is/b/2017/02/09/diving-into-dynamic-website-content-with-splashr/
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]


== Send email ==
== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
=== [https://github.com/rpremraj/mailR/ mailR] ===
See an example from [[R#RJSONIO|RJSONIO]] above.
Easiest. Require rJava package (not trivial to install, see [[#RJava|rJava]]). mailR is an interface to Apache Commons Email to send emails from within R. See also [http://unamatematicaseltigre.blogspot.com/2016/12/how-to-send-bulk-email-to-your-students.html send bulk email]


Before we use the mailR package, we have followed [https://support.google.com/accounts/answer/6010255?hl=en here] to have '''Allow less secure apps: 'ON' '''; or you might get an error ''Error: EmailException (Java): Sending the email to the following server failed : smtp.gmail.com:465''. Once we turn on this option, we may get an email for the notification of this change. Note that the recipient can be other than a gmail.
== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
<syntaxhighlight lang='rsplus'>
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
> send.mail(from = "[email protected]",
  to = c("[email protected]", "Recipient 2 <[email protected]>"),
  replyTo = c("Reply to someone else <[email protected]>")
  subject = "Subject of the email",
  body = "Body of the email",
  smtp = list(host.name = "smtp.gmail.com", port = 465, user.name = "gmail_username", passwd = "password", ssl = TRUE),
  attach.files ="./myattachment.txt",
  authenticate = TRUE,
  send = TRUE)
[1] "Java-Object{org.apache.commons.mail.SimpleEmail@7791a895}"
</syntaxhighlight>


[https://r-bar.net/mailr-smtp-webmail-starttls-tls-ssl/ MailR SMTP Setup (Gmail, Outlook, Yahoo) | STARTTLS]
== gtrendsR - Google Trends ==
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=ngs%20next%20generation%20sequencing ngs next generation sequencing] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=dna%20sequencing dna sequencing] from 2004-01-01.
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=rna%20sequencing rna sequencing] from 2004-01-01. It can be seen RNA sequencing >> DNA sequencing.
* [http://www.kdnuggets.com/2017/09/python-vs-r-data-science-machine-learning.html?utm_content=buffere1df7&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer Python vs R – Who Is Really Ahead in Data Science, Machine Learning?] and [https://stackoverflow.blog/2017/09/06/incredible-growth-python/ The Incredible Growth of Python] by [https://twitter.com/drob?lang=en David Robinson]


=== [https://cran.r-project.org/web/packages/gmailr/index.html gmailr] ===
== quantmod ==
More complicated. gmailr provides access the Google's gmail.com RESTful API. [https://cran.r-project.org/web/packages/gmailr/vignettes/sending_messages.html Vignette] and an example on [http://stackoverflow.com/questions/30144876/send-html-message-using-gmailr here]. Note that it does not use a password; it uses a '''json''' file for oauth authentication downloaded from https://console.cloud.google.com/. See also https://github.com/jimhester/gmailr/issues/1.
[http://www.thertrader.com/2015/12/13/maintaining-a-database-of-price-files-in-r/ Maintaining a database of price files in R]. It consists of 3 steps.
<syntaxhighlight lang='rsplus'>
library(gmailr)
gmail_auth('mysecret.json', scope = 'compose')


test_email <- mime() %>%
# Initial data downloading
  to("[email protected]") %>%
# Update existing data
  from("[email protected]") %>%
# Create a batch file
  subject("This is a subject") %>%
  html_body("<html><body>I wish <b>this</b> was bold</body></html>")
send_message(test_email)
</syntaxhighlight>


=== [https://cran.r-project.org/web/packages/sendmailR/index.html sendmailR] ===
== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
sendmailR provides a simple SMTP client. It is not clear how to use the package (i.e. where to enter the password).
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
* https://github.com/cran/caret source code mirrored on github
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/
* [https://daviddalpiaz.github.io/r4sl/the-caret-package.html Chapter 21 of "R for Statistical Learning"]


=== emayili ===
== Tool for connecting Excel with R ==
[https://datawookie.netlify.com/blog/2019/05/emayili-sending-email-from-r/ emayili: Sending Email from R]
* https://bert-toolkit.com/
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html


== [http://www.ncbi.nlm.nih.gov/geo/ GEO (Gene Expression Omnibus)] ==
== write.table ==
See [[GEO#R_packages|this internal link]].
=== Output a named vector ===
<pre>
vec <- c(a = 1, b = 2, c = 3)
write.csv(vec, file = "my_file.csv", quote = F)
x = read.csv("my_file.csv", row.names = 1)
vec2 <- x[, 1]
names(vec2) <- rownames(x)
all.equal(vec, vec2)


== Interactive html output ==
# one liner: row names of a 'matrix' become the names of a vector
=== [http://cran.r-project.org/web/packages/sendplot/index.html sendplot] ===
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
=== [http://cran.r-project.org/web/packages/RIGHT/index.html RIGHT] ===
all.equal(vec, vec3)
The supported plot types include scatterplot, barplot, box plot, line plot and pie plot.
</pre>


In addition to tooltip boxes, the package can create a [http://righthelp.github.io/tutorial/interactivity table showing all information about selected nodes].
=== Avoid leading empty column to header ===
[https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
<pre>
write.table(a, 'a.txt', col.names=NA)
</pre>


=== r2d3 ===
=== Add blank field AND column names in write.table ===
[https://blog.rstudio.com/2018/10/05/r2d3-r-interface-to-d3-visualizations/ r2d3] - R Interface to D3 Visualizations
* '''write.table'''(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
** Suppose x is (n x 2)
** write.table(x, sep="\t") will generate a file with 2 element on the 1st row
** read.table(file) will return an object with a size (n x 2)
** read.delim(file) and read.delim2(file) will also be correct
* Note that '''write.csv'''() does not have this issue that write.table() has
** Suppose x is (n x 2)
** Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
** If we use read.csv(file), the object is (n x 3). So we need to use '''read.csv(file, row.names = 1)'''
* adding blank field AND column names in write.table(); [https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
:<syntaxhighlight lang="rsplus">
write.table(a, 'a.txt', col.names=NA)
</syntaxhighlight>
* '''readr::write_tsv'''() does not include row names in the output file


=== [http://cran.r-project.org/web/packages/d3Network/index.html d3Network] ===
=== read.delim(, row.names=1) and write.table(, row.names=TRUE) ===
* http://christophergandrud.github.io/d3Network/ (old)
[https://www.statology.org/read-delim-in-r/ How to Use read.delim Function in R]
* https://christophergandrud.github.io/networkD3/ (new)
<source lang="rsplus">
library(d3Network)


Source <- c("A", "A", "A", "A", "B", "B", "C", "C", "D")  
Case 1: no row.names
Target <- c("B", "C", "D", "J", "E", "F", "G", "H", "I")  
<pre>
NetworkData <- data.frame(Source, Target)  
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...
</pre>
Case 2: with row.names. '''Note:''' if we open the text file in Excel, we'll see the 1st row is missing one header at the end. It is actually missing the column name for the 1st column.
<pre>
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
my_df <- read.delim('my_data.txt') # it will automatically assign the rownames
</pre>


d3SimpleNetwork(NetworkData, height = 800, width = 1024, file="tmp.html")
== Read/Write Excel files package ==
</source>
* http://www.milanor.net/blog/?p=779
 
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.  
=== [http://cran.r-project.org/web/packages/htmlwidgets/ htmlwidgets for R] ===
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
Embed widgets in R Markdown documents and Shiny web applications.
** [https://stackoverflow.com/a/17976604 Export both Image and Data from R to an Excel spreadsheet]
 
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
* Official website http://www.htmlwidgets.org/.
** It can't be used to open .xls or .xlm files.
* [http://deanattali.com/blog/htmlwidgets-tips/ How to write a useful htmlwidgets in R: tips and walk-through a real example]
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
 
** [https://fabiomarroni.wordpress.com/2018/08/07/use-r-to-write-multiple-tables-to-a-single-excel-file/ Use R to write multiple tables to a single Excel file]
=== [http://cran.r-project.org/web/packages/networkD3/index.html networkD3] ===
* [https://github.com/hadley/readxl readxl]: it does not depend on anything although it can only read but not write Excel files.
This is a port of Christopher Gandrud's [http://christophergandrud.github.io/d3Network/ d3Network] package to the htmlwidgets framework.
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
 
** [https://github.com/rstudio/webinars/tree/master/36-readxl readxl webinar].
=== [http://cran.r-project.org/web/packages/scatterD3/index.html scatterD3] ===
** One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a '''tbl_df''', a '''tbl''' and a '''data.frame.'''
scatterD3 is an HTML R widget for interactive scatter plots visualization. It is based on the htmlwidgets R package and on the d3.js javascript library.
** For writing to Excel formats, use writexl or openxlsx package.
 
:<syntaxhighlight lang='rsplus'>
=== dygraphs ===
library(readxl)
* https://github.com/rstudio/dygraphs
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE,
* [https://www.r-bloggers.com/visualizations-for-algorithmic-trading-in-r/ Visualizations for Algorithmic Trading in R]
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf,
 
    guess_max = min(1000, n_max), progress = readxl_progress(),
=== [https://github.com/bwlewis/rthreejs rthreejs] - Create interactive 3D scatter plots, network plots, and globes ===
    .name_repair = "unique")
[http://bwlewis.github.io/rthreejs/ Examples]
# Example
 
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
=== d3heatmap ===
See [[Heatmap#d3heatmap|R]]
 
=== [https://cran.r-project.org/web/packages/svgPanZoom/index.html svgPanZoom] ===
This 'htmlwidget' provides pan and zoom interactivity to R graphics, including 'base', 'lattice', and 'ggplot2'. The interactivity is provided through the 'svg-pan-zoom.js' library.
 
=== DT: An R interface to the DataTables library ===
* http://blog.rstudio.org/2015/06/24/dt-an-r-interface-to-the-datatables-library/
 
=== plotly ===
* [http://moderndata.plot.ly/power-curves-r-plotly-ggplot2/ Power curves] and ggplot2.
* [http://moderndata.plot.ly/time-series-charts-by-the-economist-in-r-using-plotly/ TIME SERIES CHARTS BY THE ECONOMIST IN R USING PLOTLY] & [https://moderndata.plot.ly/interactive-r-visualizations-with-d3-ggplot2-rstudio/ FIVE INTERACTIVE R VISUALIZATIONS WITH D3, GGPLOT2, & RSTUDIO]
* [http://moderndata.plot.ly/filled-chord-diagram-in-r-using-plotly/ Filled chord diagram]
* [https://moderndata.plot.ly/dashboards-in-r-with-shiny-plotly/ DASHBOARDS IN R WITH SHINY & PLOTLY]
* [https://plot.ly/r/shiny-tutorial/ Plotly Graphs in Shiny],
** [https://plot.ly/r/shiny-gallery/ Gallery]
** [https://plot.ly/r/shinyapp-UN-simple/ Single time series]
** [https://plot.ly/r/shinyapp-UN-advanced/ Multiple time series]
* [https://www.r-exercises.com/2017/09/28/how-to-plot-basic-charts-with-plotly/ How to plot basic charts with plotly]
* [https://www.displayr.com/how-to-add-trend-lines-in-r-using-plotly/?utm_medium=Feed&utm_source=Syndication How to add Trend Lines in R Using Plotly]
* [https://blog.methodsconsultants.com/posts/introduction-to-interactive-graphics-in-r-with-plotly/ Introduction to Interactive Graphics in R with plotly]
 
== Amazon ==
[https://github.com/56north/Rmazon Download product information and reviews from Amazon.com]
<syntaxhighlight lang='bash'>
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
</syntaxhighlight>
and in R
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R
<syntaxhighlight lang='rsplus'>
:<syntaxhighlight lang='rsplus'>
install.packages("devtools")
library(writexl)
install.packages("XML")
mylst <- list(sheet1name = df1, sheet2name = df2)
install.packages("pbapply")
write_xlsx(mylst, "output.xlsx")
install.packages("dplyr")
devtools::install_github("56north/Rmazon")
product_info <- Rmazon::get_product_info("1593273843")
reviews <- Rmazon::get_reviews("1593273843")
reviews[1,6] # only show partial characters from the 1st review
nchar(reviews[1,6])
as.character(reviews[1,6]) # show the complete text from the 1st review
 
reviews <- Rmazon::get_reviews("B07BNGJXGS")
# Fetching 30 reviews of 'BOOX Note Ereader,Android 6.0 32 GB 10.3" Dual Touch HD Display'
#  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed = 02s
reviews
# A tibble: 30 x 6
  reviewRating reviewDate reviewFormat Verified_Purcha… reviewHeadline
          <dbl> <chr>      <lgl>        <lgl>            <chr>       
1            4 May 23, 2… NA          TRUE            Good for PDF …
2            3 May 8, 20… NA          FALSE            The reading s…
3            5 May 17, 2… NA          TRUE            E-reader and …
4            3 May 24, 2… NA          TRUE            Good hardware…
5            3 June 21, … NA          TRUE            Poor QC     
6            5 August 5,… NA          TRUE            Excellent for…
7            5 May 31, 2… NA          TRUE            Especially li…
8            5 July 4, 2… NA          TRUE            Android 6 rea…
9            4 July 15, … NA          TRUE            Remember the …
10            4 June 9, 2… NA          TRUE            Overall fanta…
# ... with 20 more rows, and 1 more variable: reviewText <chr>
reviews[1, 6] # 6-th column is the review text
</syntaxhighlight>
</syntaxhighlight>


== [https://cran.r-project.org/web/packages/gutenbergr/index.html gutenbergr] ==
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).  
[https://blog.jumpingrivers.com/posts/2018/tidytext_edinbr_2018/ Edinbr: Text Mining with R]
{{Pre}}
 
> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
== Twitter ==
  UniqueID (Double-click) CloneID UGCluster
[http://www.masalmon.eu/2017/03/19/facesofr/ Faces of #rstats Twitter]
1                  HK1A1  21652 Hs.445981
 
2                  HK1A2  22012 Hs.119177
== OCR ==
3                  HK1A4  22293 Hs.501376
* [http://ropensci.org/blog/blog/2016/11/16/tesseract Tesseract package: High Quality OCR in R], [https://www.r-bloggers.com/how-to-do-optical-character-recognition-ocr-of-non-english-documents-in-r-using-tesseract/ How to do Optical Character Recognition (OCR) of non-English documents in R using Tesseract?]
                                                    Name Symbol EntrezID
* https://cran.r-project.org/web/packages/abbyyR/index.html
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1    1495
2                              ADP-ribosylation factor 3  ARF3      377
3                          Uroporphyrinogen III synthase  UROS    7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1  5.000000        5q31.2              <NA>      1
2  12.000000        12q13              <NA>      1
3      <NA> 10q25.2-q26.3              <NA>      1
</pre>


== Wikipedia ==
The hidden worksheets become visible (Not sure what are those first rows mean in the output).
[https://github.com/ironholds/wikipedir WikipediR]: R's MediaWiki API client library
{{Pre}}
> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00
[1] "Experiment descriptors" "Filtered log ratio"    "Gene identifiers"     
[4] "Gene annotations"      "CollateInfo"            "GeneSubsets"         
[7] "GeneSubsetsTemp"     
</pre>


= Creating local repository for CRAN and Bioconductor =
The Chinese character works too.
[[R_repository|R repository]]
{{Pre}}
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</pre>


= Parallel Computing =
To read all worksheets we need a convenient function
See [[R_parallel|R parallel]].
{{Pre}}
read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</pre>


= Cloud Computing =
=== [https://cran.r-project.org/web/packages/readr/ readr] ===


== Install R on Amazon EC2 ==
Compared to base equivalents like '''read.csv()''', '''readr''' is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.
http://randyzwitch.com/r-amazon-ec2/


== Bioconductor on Amazon EC2 ==
[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released. [https://www.tidyverse.org/blog/2021/07/readr-2-0-0/ readr 2.0.0] adds built-in support for reading multiple files at once, fast multi-threaded lazy reading and automatic guessing of delimiters among other changes.
http://www.bioconductor.org/help/bioconductor-cloud-ami/


= Big Data Analysis =
Consider a [http://www.cs.utoronto.ca/~juris/data/cmapbatch/instmatx.21.txt text file] where the table (6100 x 22) has duplicated row names and the (1,1) element is empty. The column names are all unique.
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
* read.delim() will treat the first column as rownames but it does not allow duplicated row names. Even we use row.names=NULL, it still does not read correctly. It does give warnings (EOF within quoted string & number of items read is not a multiple of the number of columns). The dim is 5177 x 22.
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
* readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
* [https://rstudio-pubs-static.s3.amazonaws.com/72295_692737b667614d369bd87cb0f51c9a4b.html Handling large data sets in R]
* '''data.table::fread(Filename, sep = "\t")''' will detect the number of column names is less than the number of columns. Added 1 extra default column name for the first column which is guessed to be row names or an index. The dim is 6100 x 22. (Winner!)
* [https://www.oreilly.com/library/view/big-data-analytics/9781786466457/#toc-start Big Data Analytics with R] by Simon Walkowiak
* [https://pbdr.org/publications.html pbdR]
** https://en.wikipedia.org/wiki/Programming_with_Big_Data_in_R
** [https://olcf.ornl.gov/wp-content/uploads/2016/01/pbdr.pdf Programming with Big Data in R - pbdR] George Ostrouchov and Mike Matheson Oak Ridge National Laboratory


== bigmemory, biganalytics, bigtabulate ==
The '''readr::read_csv()''' function is as fast as '''data.table::fread()''' function. ''For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv().'' See 5.3 of Efficient R Programming book.


== ff, ffbase ==
Note that '''data.table::fread()''' can read a selection of the columns.
* tapply does not work. [https://stackoverflow.com/questions/16470677/using-tapply-ave-functions-for-ff-vectors-in-r Using tapply, ave functions for ff vectors in R]
* [http://www.bnosac.be/index.php/blog/12-popularity-bigdata-large-data-packages-in-r-and-ffbase-user-presentation Popularity bigdata / large data packages in R and ffbase useR presentation]
* [http://www.bnosac.be/images/bnosac/blog/user2013_presentation_ffbase.pdf ffbase: statistical functions for large datasets] in useR 2013
* [https://www.rdocumentation.org/packages/ffbase/versions/0.12.7/topics/ffbase-package ffbase] package


== biglm ==
=== Speed comparison ===
[https://predictivehacks.com/the-fastest-way-to-read-and-write-file-in-r/ The Fastest Way To Read And Write Files In R]. data.table >> readr >> base.


== data.table ==
== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
See [[#data.table_2|data.table]].
See [[Ggplot2|ggplot2]]


= Reproducible Research =
== Data Manipulation & Tidyverse ==
http://cran.r-project.org/web/views/ReproducibleResearch.html
See [[Tidyverse|Tidyverse]].


== Reproducible Environments ==
== Data Science ==
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
See [[Data_science|Data science]] page


= Useful R packages =
== microbenchmark & rbenchmark ==
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
* [https://cran.r-project.org/web/packages/microbenchmark/index.html microbenchmark]
* [https://github.com/qinwf/awesome-R awesome-R]
** [https://www.r-bloggers.com/using-the-microbenchmark-package-to-compare-the-execution-time-of-r-expressions/ Using the microbenchmark package to compare the execution time of R expressions]
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]
* [https://cran.r-project.org/web/packages/rbenchmark/index.html rbenchmark] (not updated since 2012)


== RInside ==
== Plot, image ==
* http://dirk.eddelbuettel.com/code/rinside.html
=== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ===
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
If we want to create the image on this wiki left hand side panel, we can use the '''jpeg''' package to read an existing plot and then edit and save it.


=== Ubuntu ===
We can also use the jpeg package to import and manipulate a jpg image. See [http://moderndata.plot.ly/fun-with-heatmaps-and-plotly/ Fun with Heatmaps and Plotly].
With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on [http://www.youtube.com/watch?v=UQ8yKQcPTg0 Youtube]. I have tested this '''qtdensity''' example successfully using Qt 4.8.5.
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
# Install [[Qt|Qt]]. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
# Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
# Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the ''interactive'' program qtdensity appears on your desktop.
[[File:qtdensity.png|100px]].


With RInside + [http://www.webtoolkit.eu/wt Wt web toolkit] installed, we can also create a web application. To demonstrate the example in ''examples/wt'' directory, we can do
=== png and resolution ===
It seems people use '''res=300''' as a definition of high resolution.
 
<ul>
<li>Bottom line: fix res=300 and adjust height/width as needed. The default is res=72, height=width=480. If we increase res=300, the text font size will be increased, lines become thicker and the plot looks like a zoom-in.
<li>[https://stackoverflow.com/a/51194014 Saving high resolution plot in png].  
<pre>
<pre>
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
png("heatmap.png", width = 8, height = 6, units='in', res = 300)
make
# we can adjust width/height as we like
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
# the pixel values will be width=8*300 and height=6*300 which is equivalent to
# 8*300 * 6*300/10^6 = 4.32 Megapixels (1M pixels = 10^6 pixels) in camera's term
# However, if we use png(, width=8*300, height=6*300, units='px'), it will produce
# a plot with very large figure body and tiny text font size.
 
# It seems the following command gives the same result as above
png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
</pre>
</pre>
Then we can go to the browser's address bar and type ''http://localhost:8080'' to see how it works (a screenshot is in [http://dirk.eddelbuettel.com/blog/2011/11/30/ here]).
<li>Chapter 14.5 [https://r-graphics.org/recipe-output-bitmap Outputting to Bitmap (PNG/TIFF) Files] by R Graphics Cookbook
* Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.
<li>[https://blog.revolutionanalytics.com/2009/01/10-tips-for-making-your-r-graphics-look-their-best.html 10 tips for making your R graphics look their best] David Smith
* In Word you can resize the graphic to an appropriate size, but the high resolution gives you the flexibility to choose a size while not compromising on the quality.  I'd recommend '''at least 1200 pixels''' on the longest side for standard printers.
<li>[https://stat.ethz.ch/R-manual/R-devel/library/grDevices/html/png.html ?png]. The png function has default settings ppi=72, height=480, width=480, units="px".
* By default no resolution is recorded in the file, except for BMP.
* [https://www.adobe.com/creativecloud/file-types/image/comparison/bmp-vs-png.html BMP vs PNG format]. If you need a smaller file size and don’t mind a lossless compression, PNG might be a better choice. If you need to retain as much detail as possible and don’t mind a larger file size, BMP could be the way to go.
** '''Compression''': BMP files are raw and uncompressed, meaning they’re large files that retain as much detail as possible. On the other hand, PNG files are compressed but still lossless. This means you can reduce or expand PNGs without losing any information.
** '''File size''': BMPs are larger than PNGs. This is because PNG files automatically compress, and can be compressed again to make the file even smaller.
** '''Common uses''': BMP contains a maximum amount of details while PNGs are good for small illustrations, sketches, drawings, logos and icons.
** '''Quality''': No difference
** '''Transparency''': PNG supports transparency while BMP doesn't
<li>Some comparison about the ratio
* 11/8.5=1.29  (A4 paper)
* 8/6=1.33    (plot output)
* 1440/900=1.6 (my display)
<li>[https://babichmorrowc.github.io/post/2019-05-23-highres-figures/ Setting resolution and aspect ratios in R]
<li>The difference of '''res''' parameter for a simple plot. [https://www.tutorialspoint.com/how-to-change-the-resolution-of-a-plot-in-base-r How to change the resolution of a plot in base R?]
<li>[https://danieljhocking.wordpress.com/2013/03/12/high-resolution-figures-in-r/ High Resolution Figures in R].
<li>[https://magesblog.com/post/2013-10-29-high-resolution-graphics-with-r/ High resolution graphics with R]
<li>[https://stackoverflow.com/questions/8399100/r-plot-size-and-resolution R plot: size and resolution]
<li>[https://stackoverflow.com/a/22815896 How can I increase the resolution of my plot in R?], [https://cran.r-project.org/web/packages/devEMF/index.html devEMF] package
<li>See [[Images#Anti-alias_%E4%BF%AE%E9%82%8A|Images -> Anti-alias]].
<li>How to check DPI on PNG
* '''The width of a PNG file in terms of inches cannot be determined directly from the file itself''', as the file contains pixel dimensions, not physical dimensions. However, '''you can calculate the width in inches if you know the resolution (DPI, dots per inch) of the image'''. Remember that converting pixel measurements to physical measurements like inches involves a specific resolution (DPI), and different devices may display the same image at different sizes due to having different resolutions.
<li>[https://community.rstudio.com/t/save-high-resolution-figures-from-r-300dpi/62016/3 Cairo] case.
</ul>


=== Windows 7 ===
=== PowerPoint ===
To make RInside works on Windows OS, try the following
<ul>
# Make sure R is installed under '''C:\''' instead of '''C:\Program Files''' if we don't want to get an error like ''g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory''.
<li>For PP presentation, I found it is useful to use svg() to generate a small size figure. Then when we enlarge the plot, the text font size can be enlarged too. According to [https://www.rdocumentation.org/packages/grDevices/versions/3.6.2/topics/cairo svg], by default, width = 7, height = 7, pointsize = 12, family = '''sans'''.
# Install RTools
<li>Try the following code. The font size is the same for both plots/files. However, the first plot can be enlarged without losing its quality.
# Instal RInside package from source (the binary version will give an [http://stackoverflow.com/questions/13137770/fatal-error-unable-to-open-the-base-package error ])
# Create a DOS batch file containing necessary paths in PATH environment variable
<pre>
<pre>
@echo off
svg("svg4.svg", width=4, height=4)
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
plot(1:10, main="width=4, height=4")
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
dev.off()
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
 
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
svg("svg7.svg", width=7, height=7) # default
set R_HOME=C:\R\R-3.0.1
plot(1:10, main="width=7, height=7")
echo Setting environment for using R
dev.off()
cmd
</pre>
</pre>
In the Windows command prompt, run
</ul>
 
=== magick ===
https://cran.r-project.org/web/packages/magick/
 
See an example [[:File:Progpreg.png|here]] I created.
 
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
 
=== geDevices ===
* [https://www.jumpingrivers.com/blog/r-graphics-cairo-png-pdf-saving/ Saving R Graphics across OSs]. Use png(type="cairo-png") or the [https://cran.r-project.org/web/packages/ragg/index.html ragg] package which can be incorporated into RStudio.
* [https://www.jumpingrivers.com/blog/r-knitr-markdown-png-pdf-graphics/ Setting the Graphics Device in a RMarkdown Document]
 
=== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ===
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
 
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
<pre>
<pre>
cd C:\R\R-3.0.1\library\RInside\examples\standard
sudo apt-get install libgtk2.0-dev libcairo2-dev
make -f Makefile.win
</pre>
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
<pre>
rinside_sample0
</pre>
</pre>


As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See  some discussions in
On Windows OS, we may got the error: '''unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' '''. We need to follow the instruction in [http://tolstoy.newcastle.edu.au/R/e6/help/09/05/15613.html here].
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
* http://www.mail-archive.com/[email protected]/msg04377.html
So the Qt and Wt web tool applications on Windows may or may not be possible.


== GUI ==
=== dpi requirement for publication ===
=== Qt and R ===
[http://www.cookbook-r.com/Graphs/Output_to_a_file/ For import into PDF-incapable programs (MS Office)]
* http://cran.r-project.org/web/packages/qtbase/index.html [https://stat.ethz.ch/pipermail/r-devel/2015-July/071495.html QtDesigner is such a tool, and its output is compatible with the qtbase R package]
* http://qtinterfaces.r-forge.r-project.org


== tkrplot ==
=== sketcher: photo to sketch effects ===
On Ubuntu, we need to install tk packages, such as by
https://htsuda.net/sketcher/
<pre>
sudo apt-get install tk-dev
</pre>


== reticulate - Interface to 'Python' ==
=== httpgd ===
* https://cran.r-project.org/web/packages/reticulate/index.html, [https://github.com/rstudio/reticulate Github]
* https://nx10.github.io/httpgd/ A graphics device for R that is accessible via network protocols. Display graphics on browsers.
** Using Python in R markdown
* [https://youtu.be/uxyhmhRVOfw Three tricks to make IDEs other than Rstudio better for R development]
** Importing Python modules and call its functions directly from R — '''import()''' function
** Sourcing Python scripts — '''source_python()''' function
** Python REPL — The '''repl_python()''' function creates an interactive Python console within R.
* Install Python packages https://rstudio.github.io/reticulate/articles/python_packages.html
** Better to have [https://www.anaconda.com/distribution/ anaconda3] installed. 2.26G space is required on macOS.
** Direct running py_install("pandas") would ask me to upgrade virtualenv
** Running virtualenv_create("r-reticulate") and then py_install("pandas") works
* [https://blog.rstudio.com/2018/03/26/reticulate-r-interface-to-python/ reticulate: R interface to Python] JJ Allaire
* [https://www.rstudio.com/resources/cheatsheets/ Cheat sheet]
* [https://www.brodrigues.co/blog/2018-12-30-reticulate/ R or Python? Why not both? Using Anaconda Python within R with {reticulate}]
* [https://www.listendata.com/2018/03/run-python-from-r.html?m=1 Run Python from R]
* [https://www.statworx.com/de/blog/r-and-python-using-reticulate-to-get-the-best-of-both-worlds/ R and Python: Using reticulate to get the best of both worlds]. Note
** [https://rstudio.github.io/reticulate/articles/r_markdown.html RStudio v1.2 preview release includes support for using reticulate to execute Python chunks within R Notebooks]
** Error from my execution: ''ValueError: 'RBF' is not in list''
* [https://rviews.rstudio.com/2019/03/18/the-reticulate-package-solves-the-hardest-problem-in-data-science-people/ The reticulate package solves the hardest problem in data science: people]
* [https://rviews.rstudio.com/2019/06/10/reticulate-virtualenv-and-python-in-linux/ reticulate, virtualenv, and Python in Linux]
* Bugs
** [https://stackoverflow.com/a/49556037 Pass Python objects to R]: Works. Or use py_run_string()
** [https://stackoverflow.com/a/52542230 Cannot pass R variables to Python]: use source_python()
* [https://github.com/matloff/R-vs.-Python-for-Data-Science R vs Python for data science] by Norm Matloff.
* Test python and markdown files
<syntaxhighlight lang='python'>
def add_three(x):
    z = x + 3
    return z
</syntaxhighlight>


<pre>
== [http://igraph.org/r/ igraph] ==
---
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]
title: "R Notebook"
output: html_notebook
---


```{r}
== Identifying dependencies of R functions and scripts ==
library(reticulate)
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
py_discover_config()
{{Pre}}
x <- 5
library(mvbutils)
source_python("test.py")
foodweb(where = "package:batr")
y <- add_three(x)
print(y)
```


Pass R variables to Python. Works
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
```{python}
a = 7
print(r.x)
```


Pass python variables to R. Works.
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
```{r}
py$a
py_run_string("y = 10"); py$y
```
</pre>
</pre>


== Hadoop (eg ~100 terabytes) ==
== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
Iterator is useful over for-loop if the data is already a '''collection'''. It can be used to iterate over a vector, data frame, matrix, file


* RHadoop
Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
* Hive
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].
* http://www.techspritz.com/category/tutorials/hadoopmapredcue/ Single node or multinode cluster setup using Ubuntu with VirtualBox (Excellent)
* [http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/ Running Hadoop on Ubuntu Linux (Single-Node Cluster)]
* Ubuntu 12.04 http://www.youtube.com/watch?v=WN2tJk_oL6E and [https://www.dropbox.com/s/05aurcp42asuktp/Chiu%20Hadoop%20Pig%20Install%20Instructions.docx instruction]
* Linux Mint http://blog.hackedexistence.com/installing-hadoop-single-node-on-linux-mint
* http://www.r-bloggers.com/search/hadoop


=== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ===
== Colors ==
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
* Ubuntu 12.04 - [http://crishantha.com/wp/?p=1414 Crishantha.com], [http://nikhilshah123sh.blogspot.com/2014/03/setting-up-rhadoop-in-ubuntu-1204.html nikhilshah123sh.blogspot.com].[http://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/ Bighadoop.wordpress] contains an example.
<ul>
* RapReduce in R by [https://github.com/RevolutionAnalytics/rmr2/blob/master/docs/tutorial.md RevolutionAnalytics] with a few examples.
<li>[https://cran.r-project.org/web/packages/colorspace/index.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes. Popular! Many reverse imports/suggests; e.g. ComplexHeatmap. See my [[Ggplot2#colorspace_package|ggplot2]] page.
* https://twitter.com/hashtag/rhadoop
<pre>
* [http://bigd8ta.com/step-by-step-guide-to-setting-up-an-r-hadoop-system/ Bigd8ta.com] based on Ubuntu 14.04.
hcl_palettes(plot = TRUE) # a quick overview
hcl_palettes(palette = "Dark 2", n=5, plot = T)
q4 <- qualitative_hcl(4, palette = "Dark 3")
</pre>
</ul>
* [https://statisticsglobe.com/create-color-range-between-two-colors-in-r Create color range between two colors in R] using colorRampPalette()
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* palette_explorer() function from the [https://cran.r-project.org/web/packages/tmaptools/index.html tmaptools] package. See [https://www.computerworld.com/article/3184778/data-analytics/6-useful-r-functions-you-might-not-know.html selecting color palettes with shiny].
* [http://www.cookbook-r.com/ Cookbook for R]
* [http://ggplot2.tidyverse.org/reference/scale_brewer.html Sequential, diverging and qualitative colour scales/palettes from colorbrewer.org]: scale_colour_brewer(), scale_fill_brewer(), ...
* http://colorbrewer2.org/
* It seems there is no choice of getting only 2 colors no matter which set name we can use
* To see the set names used in brewer.pal, see
** [https://www.rdocumentation.org/packages/RColorBrewer/versions/1.1-2/topics/RColorBrewer RColorBrewer::display.brewer.all()]
** [https://rpubs.com/flowertear/224344 Output]
** Especially, '''[http://colorbrewer2.org/#type=qualitative&scheme=Set1&n=4 Set1]''' from http://colorbrewer2.org/
* To list all R color names, colors().
** [http://research.stowers.org/mcm/efg/R/Color/Chart/ColorChart.pdf Color Chart] (include Hex and RGB) & [http://research.stowers.org/mcm/efg/Report/UsingColorInR.pdf Using Color in R] from http://research.stowers.org
** Code to generate rectangles with colored background https://www.r-graph-gallery.com/42-colors-names/
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
* http://deanattali.com/blog/colourpicker-package/ Not sure what it is doing
* [http://www.lifehack.org/484519/how-to-choose-the-best-colors-for-your-data-charts How to Choose the Best Colors For Your Data Charts]
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* [http://sape.inf.usi.ch/quick-reference/ggplot2/colour Color names in R]
<ul>
<li>[https://stackoverflow.com/questions/28461326/convert-hex-color-code-to-color-name convert hex value to color names]
{{Pre}}
library(plotrix)
sapply(rainbow(4), color.id) # color.id is a function
          # it is used to identify closest match to a color
sapply(palette(), color.id)
sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
</pre>
</li></ul>
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.3/topics/hsv hsv()] function. [https://eranraviv.com/matrix-style-screensaver-in-r/ Matrix-style screensaver in R]


=== Snowdoop: an alternative to MapReduce algorithm ===
Below is an example using the option ''scale_fill_brewer''(palette = "[http://colorbrewer2.org/#type=qualitative&scheme=Paired&n=9 Paired]"). See the source code at [https://gist.github.com/JohannesFriedrich/c7d80b4e47b3331681cab8e9e7a46e17 gist]. Note that only '''set1''' and '''set3''' palettes in '''qualitative scheme''' can support up to 12 classes.
* http://matloff.wordpress.com/2014/11/26/how-about-a-snowdoop-package/
* http://matloff.wordpress.com/2014/12/26/snowdooppartools-update/comment-page-1/#comment-665


== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
On Ubuntu, we need to install libxml2-dev before we can install XML package.
<pre>
sudo apt-get update
sudo apt-get install libxml2-dev
</pre>


On CentOS,
[[:File:GgplotPalette.svg]]
<pre>
yum -y install libxml2 libxml2-devel
</pre>


=== XML ===
=== [http://rpubs.com/gaston/colortools colortools] ===
* http://giventhedata.blogspot.com/2012/06/r-and-web-for-beginners-part-ii-xml-in.html. It gave an example of extracting the XML-values from each XML-tag for all nodes and save them in a data frame using '''xmlSApply()'''.
Tools that allow users generate color schemes and palettes
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
* https://www.tutorialspoint.com/r/r_xml_files.htm
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
* [http://www.stat.berkeley.edu/~statcur/Workshop2/Presentations/XML.pdf Extracting data from XML] PubMed and Zillow are used to illustrate. xmlTreeParse(),  xmlRoot(),  xmlName() and xmlSApply().
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
<syntaxhighlight lang='rsplus'>
library(XML)


# Read and parse HTML file
=== [https://github.com/daattali/colourpicker colourpicker] ===
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)
A Colour Picker Tool for Shiny and for Selecting Colours in Plots


# Extract all the paragraphs (HTML tag is p, starting at
=== eyedroppeR ===
# the root of the document). Unlist flattens the list to
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


# Replace all by spaces
== [https://github.com/kevinushey/rex rex] ==
doc.text = gsub('\n', ' ', doc.text)
Friendly Regular Expressions


# Join all the elements of the character vector into a single
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
# character string, separated by spaces
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
doc.text = paste(doc.text, collapse = ' ')
</syntaxhighlight>


This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
<syntaxhighlight lang='rsplus'>
> library(RCurl) # getURL()
> library(XML)  # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # <span class="css-truncate-target">2.5.3a</span>
> plain.text
[1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"   
[6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"


> # try picard
<pre>
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
library(formatR)
> doc = htmlParse(xData)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
tidy_source("clipboard")  
[1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
# default width is getOption("width") which is 127 in my case.
[10] "2.6.0"
</pre>
</syntaxhighlight>
This method can be used to monitor new tags/releases from some projects like [https://github.com/Ultimaker/Cura/releases Cura], BWA, Picard, [https://github.com/alexdobin/STAR/releases STAR]. But for some projects like [https://github.com/ncbi/sra-tools sratools] the '''class''' attribute in the '''span''' element ("css-truncate-target") can be different (such as "tag-name").


=== xmlview ===
Some issues
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
 
== RCurl ==
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
<syntaxhighlight lang='bash'>
# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
 
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R
 
No google ID is required
 
Seems not work
<pre>
<pre>
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']", :
cat("abcd",
  arguments imply differing number of rows: 2, 0
    # This is my comment
    "defg")
</pre>
</pre>
 
will result in
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
<pre>
'''devtools''' package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the [[#remotes|remotes]] package which was suggested by the [https://cran.r-project.org/web/packages/BiocManager/index.html BiocManager] package.
> tidy_source("clipboard")
<syntaxhighlight lang='bash'>
Error in base::parse(text = code, srcfile = NULL) :
# Ubuntu 14.04
  3:1: unexpected string constant
sudo apt-get install libcurl4-openssl-dev
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
 
3: "defg"
# Ubuntu 16.04
  ^
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev
</pre>
</syntaxhighlight>
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.  
 
<pre>
[https://github.com/wch/movies/issues/3 Lazy-load database XXX is corrupt. internal error -3]. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.
cat("abcd"
 
    ,"defg"  # This is my comment
=== [https://github.com/hadley/httr httr] ===
  ,"ghij")
httr imports curl, jsonlite, mime, openssl and R6 packages.
</pre>
 
will become
When I tried to install httr package, I got an error and some message:
<pre>
<pre>
Configuration failed because openssl was not found. Try installing:
cat("abcd", "defg"  # This is my comment
* deb: libssl-dev (Debian, Ubuntu, etc)
, "ghij")
* rpm: openssl-devel (Fedora, CentOS, RHEL)
</pre>
* csw: libssl_dev (Solaris)
Still bad!!
* brew: openssl (Mac OSX)
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
If openssl is already installed, check that 'pkg-config' is in your
<pre>
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
cat("</p>",
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
--------------------------------------------------------------------
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),
ERROR: configuration failed for package ‘openssl’
file=ExternalFileName, sep="\n", append=T)
</pre>
will result in
<pre>
> tidy_source("clipboard", width.cutoff=70)
Error in base::parse(text = code, srcfile = NULL) :
  3:129: unexpected SPECIAL
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
</pre>
* ''width.cutoff'' parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
<pre>
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,
    sep = "\n", append = T)
</pre>
* It merges lines though I don't always want to do that. For example
<pre>
cat("abcd"
    ,"defg" 
  ,"ghij")
</pre>
will become
<pre>
cat("abcd", "defg", "ghij")
</pre>
</pre>
It turns out after I run '''sudo apt-get install libssl-dev''' in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!
Real example: see [http://stackoverflow.com/questions/27371372/httr-retrieving-data-with-post this post]. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).
Since httr package was used in many other packages, take a look at how others use it. For example, [https://github.com/ropensci/aRxiv aRxiv] package.
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
curl is independent of RCurl package.
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
* https://www.opencpu.org/posts/curl-release-0-8/


<syntaxhighlight lang='rsplus'>
== styler ==
library(curl)
https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)
</syntaxhighlight>


=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
== Download papers ==
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
Search and Download Papers from the bioRxiv Preprint Server (biology)


== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).
Interface to the arXiv API


Example:
=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
<syntaxhighlight lang='rsplus'>
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
# https://github.com/henrikbengtsson/matrixstats
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
remotes::install_github('HenrikBengtsson/matrixStats@develop')
* https://ropensci.org/technotes/2018/12/14/pdftools-20/
</syntaxhighlight>


== DirichletMultinomial ==
== [https://github.com/ColinFay/aside aside]: set it aside ==
On Ubuntu, we do
An RStudio addin to run long R commands aside your current session.
<pre>
sudo apt-get install libgsl0-dev
</pre>


== Create GUI ==
== Teaching ==
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics


== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
== Organize R research project ==
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]
* [https://cran.r-project.org/web/views/ReproducibleResearch.html CRAN Task View: Reproducible Research]
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
* [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow]. It suggests the [https://github.com/r-lib/here here] package. Don't use '''setwd()''' and '''rm(list = ls())'''.
** [https://rstats.wtf/safe-paths.html Practice safe paths]. Use projects and the [https://cran.r-project.org/web/packages/here/index.html here] package.
** In RStudio, if we try to send a few lines of code and one of the line contains '''setwd()''', it will give a message: ''The working directory was changed to XXX inside a notebook chunk. The working directory will be reset when the chunk is finished running. Use the knitr root.dir option in the setup chunk to change the working directory for notebook chunks.''
** [http://jenrichmond.rbind.io/post/how-to-use-the-here-package/ how to use the `here` package]
** No update for the ''here'' package after 2020-12. Consider [https://github.com/r-lib/usethis usethis] package (Automate project and package setup).
* drake project
** [https://ropensci.org/blog/2018/02/06/drake/ The prequel to the drake R package]
** [https://ropenscilabs.github.io/drake-manual/index.html The drake R Package User Manual]
* [https://docs.ropensci.org/targets/ targets] package
* [http://projecttemplate.net/ ProjectTemplate]


== [http://cran.r-project.org/web/packages/rjson/index.html rjson] ==
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/
[https://rcrastinate.rbind.io/post/how-to-save-and-load-data-in-r-an-overview/ How to save (and load) datasets in R: An overview]


== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ==
=== Naming convention ===
=== Accessing Bitcoin Data with R ===
<ul>
http://blog.revolutionanalytics.com/2015/11/accessing-bitcoin-data-with-r.html
<li>[https://stackoverflow.com/a/1946879 What is your preferred style for naming variables in R?]
* Use of period separator: they can get mixed up in simple method dispatch. However, it is used by base R ([https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/make.names make.names()], read.table(), et al)
* Use of underscores: really annoying for ESS users
* '''camelCase''': Winner
<li>However, the [https://stackoverflow.com/a/13413278 survey] said (no surprises perhaps) that
* '''lowerCamelCase''' was most often used for function names,
* '''period.separated''' names most often used for parameters.
<li>[https://datamanagement.hms.harvard.edu/collect/file-naming-conventions What are file naming conventions?]
<li>[https://www.r-bloggers.com/2014/07/consistent-naming-conventions-in-r/ Consistent naming conventions in R]
<li>http://adv-r.had.co.nz/Style.html
<li>[https://www.r-bloggers.com/2011/07/testing-for-valid-variable-names/ Testing for valid variable names]
<li>R reserved words ?Reserved
* [https://www.datamentor.io/r-programming/reserved-words/ R Reserved Words]
* Among these words, if, else, repeat, while, function, for, '''in''', next and break are used for conditions, loops and user defined functions.
<li>Microarray/RNA-seq data
<pre>
clinicalDesignData  # clnDesignData
geneExpressionData  # gExpData
geneAnnotationData  # gAnnoData


=== Plot IP on google map ===
dataClinicalDesign
* http://thebiobucket.blogspot.com/2011/12/some-fun-with-googlevis-plotting-blog.html#more  (RCurl, RJONIO, plyr, googleVis)
dataGeneExpression
* http://devblog.icans-gmbh.com/using-the-maxmind-geoip-api-with-r/ (RCurl, RJONIO, maps)
dataAnnotation
* http://cran.r-project.org/web/packages/geoPlot/index.html (geoPlot package (deprecated as 8/12/2013))
</pre>
* http://archive09.linux.com/feature/135384  (Not R) ApacheMap
* http://batchgeo.com/features/geolocation-ip-lookup/    (Not R)  (Enter a spreadsheet of adress, city, zip or a column of IPs and it will show the location on google map)
* http://code.google.com/p/apachegeomap/
 
The following example is modified from the first of above list.
<pre>
<pre>
require(RJSONIO) # fromJSON
# Search all variables ending with .Data
require(RCurl)   # getURL
ls()[grep("\\.Data$", ls())]
# Search all variables starting with data_
ls()[grep("^data_", ls())]
</pre>
</ul>


temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
=== Efficient Data Management in R ===
                          ssl.verifypeer = FALSE)
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.
ip <- read.table(textConnection(temp), as.is=TRUE)
names(ip) <- "IP"
nr = nrow(ip)
Lon <- as.numeric(rep(NA, nr))
Lat <- Lon
Coords <- data.frame(Lon, Lat)
ip2coordinates <- function(ip) {
  api <- "http://freegeoip.net/json/"
  get.ips <- getURL(paste(api, URLencode(ip), sep=""))
  # result <- ldply(fromJSON(get.ips), data.frame)
  result <- data.frame(fromJSON(get.ips))
  names(result)[1] <- "ip.address"
  return(result)
}


for (i in 1:nr){
== Text to speech ==
  cat(i, "\n")
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]
  try(
  Coords[i, 1:2] <- ip2coordinates(ip$IP[i])[c("longitude", "latitude")]
  )
}
# append to log-file:
logfile <- data.frame(ip, Lat = Coords$Lat, Long = Coords$Lon,
                                      LatLong = paste(round(Coords$Lat, 1), round(Coords$Lon, 1), sep = ":"))
log_gmap <- logfile[!is.na(logfile$Lat), ]


require(googleVis) # gvisMap
== Speech to text ==
gmap <- gvisMap(log_gmap, "LatLong",
https://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]
                options = list(showTip = TRUE, enableScrollWheel = TRUE,
                              mapType = 'hybrid', useMapTypeControl = TRUE,
                              width = 1024, height = 800))
plot(gmap)
</pre>
[[File:GoogleVis.png|200px]]


The plot.gvis() method in googleVis packages also teaches the startDynamicHelp() function in the tools package, which was used to launch a http server. See
== Weather data ==
[http://jeffreyhorner.tumblr.com/page/3 Jeffrey Horner's note about deploying Rook App].
* [https://github.com/ropensci/prism prism] package
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== Map ==
== logR ==
=== [https://rstudio.github.io/leaflet/ leaflet] ===
https://github.com/jangorecki/logR
* rstudio.github.io/leaflet/#installation-and-use
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


=== choroplethr ===
== Progress bar ==
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
https://github.com/r-lib/progress#readme
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/


=== ggplot2 ===
Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]


== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
== cron ==
See an example from [[R#RJSONIO|RJSONIO]] above.
* [https://github.com/bnosac/cronr cronR]
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]


== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
== beepr: Play A Short Sound ==
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".


== gtrendsR - Google Trends ==
== utils package ==
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
https://www.rdocumentation.org/packages/utils/versions/3.6.2
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=ngs%20next%20generation%20sequencing ngs next generation sequencing] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=dna%20sequencing dna sequencing] from 2004-01-01.
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=rna%20sequencing rna sequencing] from 2004-01-01. It can be seen RNA sequencing >> DNA sequencing.
* [http://www.kdnuggets.com/2017/09/python-vs-r-data-science-machine-learning.html?utm_content=buffere1df7&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer Python vs R – Who Is Really Ahead in Data Science, Machine Learning?] and [https://stackoverflow.blog/2017/09/06/incredible-growth-python/ The Incredible Growth of Python] by [https://twitter.com/drob?lang=en David Robinson]


== quantmod ==
== tools package ==
[http://www.thertrader.com/2015/12/13/maintaining-a-database-of-price-files-in-r/ Maintaining a database of price files in R]. It consists of 3 steps.
* https://www.rdocumentation.org/packages/tools/versions/3.6.2
* [https://www.r-bloggers.com/2023/08/three-four-r-functions-i-enjoyed-this-week/ Where in the file are there non ASCII characters?], [https://rdocumentation.org/packages/tools/versions/3.6.2/topics/showNonASCII tools::showNonASCIIfile(<filename>)]


# Initial data downloading
= Different ways of using R =
# Update existing data
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
# Create a batch file


== [http://cran.r-project.org/web/packages/Rcpp/index.html Rcpp] ==
== 10 things R can do that might surprise you ==
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/


* [http://lists.r-forge.r-project.org/pipermail/rcpp-devel/ Discussion archive]
== R call C/C++ ==
* (Video) [https://www.rstudio.com/resources/videos/extending-r-with-c-a-brief-introduction-to-rcpp/ Extending R with C++: A Brief Introduction to Rcpp]
Mainly talks about .C() and .Call().
* [http://dirk.eddelbuettel.com/blog/2017/06/13/#007_c++14_r_travis C++14, R and Travis -- A useful hack]


It may be necessary to install dependency packages for RcppEigen.
Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.
<syntaxhighlight lang='rsplus'>
sudo apt-get install libblas-dev liblapack-dev
sudo apt-get install gfortran
</syntaxhighlight>


=== Speed Comparison ===
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
* [http://blog.revolutionanalytics.com/2015/06/a-comparison-of-high-performance-computing-techniques-in-r.html A comparison of high-performance computing techniques in R]. It compares Rcpp to an R looping operator (like mapply), a parallelized version of a looping operator (like mcmapply), explicit parallelization, via the parallel package or the ParallelR suite.
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
* In the following example, C++ avoids the overhead of creating an intermediate object (eg vector of the same length as the original vector). The c++ uses an intermediate scalar. So C++ wins R over memory management in this case. <syntaxhighlight lang='rsplus'>
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
# http://blog.mckuhn.de/2016/03/avoiding-unnecessary-memory-allocations.html  
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
library(Rcpp)
* http://www.stat.harvard.edu/ccr2005/
* http://mazamascience.com/WorkingWithData/?p=1099
* [https://youtube.com/playlist?list=PLwc48KSH3D1OkObQ22NHbFwEzof2CguJJ Make an R package with C++ code] (a playlist from youtube)
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-calling-c-code-hello-world/ Using R – Calling C code ‘Hello World!’]
* [http://www.haowulab.org//pages/computing.html Computing tip] by Hao Wu


`%count<%` <- cppFunction('
=== .Call ===
size_t count_less(NumericVector x, NumericVector y) {
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/CallExternal ?.Call]
  const size_t nx = x.size();
* [http://mazamascience.com/WorkingWithData/?p=1099 Using R — .Call(“hello”)]
  const size_t ny = y.size();
* http://adv-r.had.co.nz/C-interface.html
  if (nx > 1 & ny > 1) stop("Only one parameter can be a vector!");
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-callhello/ Using R – .Call(“hello”)]
  size_t count = 0;
  if (nx == 1) {
    double c = x[0];
    for (int i = 0; i < ny; i++) count += c < y[i];
  } else {
    double c = y[0];
    for (int i = 0; i < nx; i++) count += x[i] < c;
  }
  return count;
}
')


set.seed(42)
Be sure to add the ''PACKAGE'' parameter to avoid an error like
<pre>
cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group,
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) :
  "standardize" not resolved from current namespace (grpreg)
</pre>


N <- 10^7
=== NAMESPACE file & useDynLib ===
v <- runif(N, 0, 10000)
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
* We don't need to include double quotes around the C/Fortran subroutines in .C() or .Fortran()
* digest package example: [https://github.com/cran/digest/blob/master/NAMESPACE NAMESPACE] and [https://github.com/cran/digest/blob/master/R/digest.R R functions] using .Call().
* stats example: [https://github.com/wch/r-source/blob/trunk/src/library/stats/NAMESPACE NAMESPACE]


# Testing on my ODroid xu4 running ubuntu 15.10
(From [https://cran.r-project.org/doc/manuals/r-release/R-exts.html#dyn_002eload-and-dyn_002eunload Writing R Extensions manual]) Loading is most often done automatically based on the '''useDynLib()''' declaration in the '''NAMESPACE''' file, but may be done explicitly via a call to '''library.dynam()'''. This has the form
system.time(sum(v < 5000))
{{Pre}}
#  user  system elapsed
library.dynam("libname", package, lib.loc)  
#  1.135  0.305  1.453
</pre>
system.time(v %count<% 5000)
#  user  system elapsed
#  0.535  0.000  0.540
</syntaxhighlight>
* [http://blog.ephorie.de/why-r-for-data-science-and-not-python Why R for data science – and not Python?]<syntaxhighlight lang='rsplus'>
library(Rcpp)
bmi_R <- function(weight, height) {
  weight / (height * height)
}
bmi_R(80, 1.85) # body mass index of person with 80 kg and 185 cm
## [1] 23.37473
cppFunction("
  float bmi_cpp(float weight, float height) {
    return weight / (height * height);
  }
")
bmi_cpp(80, 1.85) # same with cpp function
## [1] 23.37473
</syntaxhighlight>
* [https://www.enchufa2.es/archives/boost-the-speed-of-r-calls-from-rcpp.html Boost the speed of R calls from Rcpp]


=== Use Rcpp in RStudio ===
=== library.dynam.unload() ===
RStudio makes it easy to use Rcpp package.
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
* http://r-pkgs.had.co.nz/src.html. The '''library.dynam.unload()''' function should be placed in '''.onUnload()''' function. This function can be saved in any R files.
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]


Open RStudio, click New File -> C++ File. It will create a C++ template on the RStudio editor
=== gcc ===
<pre>
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
#include <Rcpp.h>
using namespace Rcpp;


// Below is a simple example of exporting a C++ function to R. You can
=== Primitive functions ===
// source this function into an R session using the Rcpp::sourceCpp
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]
// function (or via the Source button on the editor toolbar)


// For more on using Rcpp click the Help button on the editor toolbar
== SEXP ==
Some examples from packages


// [[Rcpp::export]]
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
int timesTwo(int x) {
  return x * 2;
}
</pre>
Now in R console, type
<pre>
library(Rcpp)
sourceCpp("~/Downloads/timesTwo.cpp")
timesTwo(9)
# [1] 18
</pre>
See more examples on http://adv-r.had.co.nz/Rcpp.html and [http://blog.revolutionanalytics.com/2017/08/kmeans-r-rcpp.html Calculating a fuzzy kmeans membership matrix]


If we wan to test Boost library, we can try it in RStudio. Consider the following example in [http://stackoverflow.com/questions/19034564/can-the-bh-r-package-link-to-boost-math-and-numeric stackoverflow.com].
== R call Fortran ==
<pre>
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
// [[Rcpp::depends(BH)]]
* [https://www.r-bloggers.com/the-need-for-speed-part-1-building-an-r-package-with-fortran-or-c/ The Need for Speed Part 1: Building an R Package with Fortran (or C)] (Very detailed)
#include <Rcpp.h>
#include <boost/foreach.hpp>
#include <boost/math/special_functions/gamma.hpp>


#define foreach BOOST_FOREACH
== Embedding R ==


using namespace boost::math;
* See [http://cran.r-project.org/doc/manuals/R-exts.html#Linking-GUIs-and-other-front_002dends-to-R Writing for R Extensions] Manual Chapter 8.
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report]  by Friedrich Leisch in 2007.
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl


//[[Rcpp::export]]
=== An very simple example (do not return from shell) from Writing R Extensions manual ===
Rcpp::NumericVector boost_gamma( Rcpp::NumericVector x ) {
The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.
  foreach( double& elem, x ) {
    elem = boost::math::tgamma(elem);
  };


  return x;
This example can be run by
}
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
</pre>
 
Then the R console
Note:  
<pre>
# '''R_HOME/bin/exec/R''' is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
boost_gamma(0:10 + 1)
# '''R_HOME/bin/R''' is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to ''/usr/local/bin/R''. When we run ''R_HOME/bin/R'', it actually runs ''R_HOME/bin/R CMD R_HOME/bin/exec/R'' (see line 259 of ''R_HOME/bin/R'' as in R 3.0.2) so we know the important role of ''R_HOME/bin/exec/R''.
# [1]      1      1      2      6      24    120    720    5040  40320
# [10]  362880 3628800


identical( boost_gamma(0:10 + 1), factorial(0:10) )
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
# [1] TRUE
</pre>


=== Example 1. convolution example ===
=== An example from Bioconductor workshop ===
First, Rcpp package should be installed (I am working on Linux system). Next we try one example shipped in Rcpp package.
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c


PS. If R was not available in global environment (such as built by ourselves), we need to modify 'Makefile' file by replacing 'R' command with its complete path (4 places).
Example:
Create [https://gist.github.com/arraytools/7d32d92fee88ffc029365d178bc09e75#file-embed-c embed.c] file.
Then build the executable. Note that I don't need to create R_HOME variable.
<pre>
<pre>
cd ~/R/x86_64-pc-linux-gnu-library/3.0/Rcpp/examples/ConvolveBenchmarks/
cd  
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
make
R
~/R-3.0.1/bin/R CMD ./Rtest
</pre>
Then type the following in an R session to see how it works. Note that we don't need to issue '''library(Rcpp)''' in R.
<pre>
dyn.load("convolve3_cpp.so")
x <- .Call("convolve3cpp", 1:3, 4:6)
x # 4 13 28 27 18
</pre>


If we have our own cpp file, we need to use the following way to create dynamic loaded library file. Note that the  character ([http://bash.cyberciti.biz/guide/Command_substitution grave accent]) ` is not (single quote)'. If you mistakenly use ', it won't work.
nano embed.c
<pre>
# Using a single line will give an error and cannot not show the real problem.
export PKG_CXXFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
export PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
# A better way is to run compile and link separately
R CMD SHLIB xxxx.cpp
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed
</pre>
</pre>


=== Example 2. Use together with inline package ===
Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
* http://adv-r.had.co.nz/C-interface.html#calling-c-functions-from-r
<pre>
<pre>
library(inline)
export R_HOME=/home/brb/Downloads/R-3.0.2
src <-'
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
Rcpp::NumericVector xa(a);
./embed # No need to include R CMD in front.
Rcpp::NumericVector xb(b);
int n_xa = xa.size(), n_xb = xb.size();
 
Rcpp::NumericVector xab(n_xa + n_xb - 1);
for (int i = 0; i < n_xa; i++)
for (int j = 0; j < n_xb; j++)
xab[i + j] += xa[i] * xb[j];
return xab;
'
fun <- cxxfunction(signature(a = "numeric", b = "numeric"),
src, plugin = "Rcpp")
fun(1:3, 1:4)
# [1]  1  4 10 16 17 12
</pre>
</pre>


=== Example 3. Calling an R function ===
Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].


=== [http://cran.r-project.org/web/packages/RcppParallel/index.html RcppParallel] ===
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf


=== rTRNG: parallel RNG in R ===
=== Create a Simple Socket Server in R ===
[https://mirai-solutions.ch/news/2019/06/10/rTRNG-avanced-parallel-RNG-R/ rTRNG: ADVANCED PARALLEL RNG IN R]
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].


== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
Create an R function
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
<pre>
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
simpleServer <- function(port=6543)
* https://github.com/cran/caret source code mirrored on github
{
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.


== Tool for connecting Excel with R ==
Welcome to R!
* https://bert-toolkit.com/
R> summary(iris[, 3:5])
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
  Petal.Length    Petal.Width          Species 
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html
Min.  :1.000  Min.  :0.100  setosa    :50 
1st Qu.:1.600  1st Qu.:0.300  versicolor:50 
Median :4.350  Median :1.300  virginica :50 
Mean  :3.758  Mean  :1.199                 
3rd Qu.:5.100  3rd Qu.:1.800                 
Max.  :6.900  Max.   :2.500                 


== Read/Write Excel files package ==
R> quit
* http://www.milanor.net/blog/?p=779
Connection closed by foreign host.
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
</pre>
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
** It can't be used to open .xls or .xlm files.
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
** [https://fabiomarroni.wordpress.com/2018/08/07/use-r-to-write-multiple-tables-to-a-single-excel-file/ Use R to write multiple tables to a single Excel file]
* [https://github.com/hadley/readxl readxl]: it does not depend on anything although it can only read but not write Excel files. 
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
** [https://github.com/rstudio/webinars/tree/master/36-readxl readxl webinar].
** One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a '''tbl_df''', a '''tbl''' and a '''data.frame.'''
** For writing to Excel formats, use writexl or openxlsx package.
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R


<syntaxhighlight lang='rsplus'>
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
library(readxl)
Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See [[R#An_example_from_Bioconductor_workshop|Example from Bioconductor workshop]].
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE,
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf,
    guess_max = min(1000, n_max), progress = readxl_progress(),
    .name_repair = "unique")
# Example
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
</syntaxhighlight>
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).  
<syntaxhighlight lang='rsplus'>
> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                  HK1A1  21652 Hs.445981
2                  HK1A2  22012 Hs.119177
3                  HK1A4  22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1    1495
2                              ADP-ribosylation factor 3  ARF3      377
3                          Uroporphyrinogen III synthase  UROS    7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1  5.000000        5q31.2              <NA>      1
2  12.000000        12q13              <NA>      1
3      <NA> 10q25.2-q26.3              <NA>      1
</syntaxhighlight>


The hidden worksheets become visible (Not sure what are those first rows mean in the output).
See my [[Rserve]] page.
<syntaxhighlight lang='rsplus'>
> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00
[1] "Experiment descriptors" "Filtered log ratio"    "Gene identifiers"     
[4] "Gene annotations"      "CollateInfo"            "GeneSubsets"         
[7] "GeneSubsetsTemp"     
</syntaxhighlight>


The Chinese character works too.
=== outsider ===
<syntaxhighlight lang='rsplus'>
* [https://joss.theoj.org/papers/10.21105/joss.02038 outsider]: Install and run programs, outside of R, inside of R
> read_excel("~/Downloads/testChinese.xlsx", 1)
* [https://github.com/stephenturner/om..bcftools Run bcftools with outsider in R]
  中文 B C
1    a b c
2    1 2 3
</syntaxhighlight>


To read all worksheets we need a convenient function
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
<syntaxhighlight lang='rsplus'>
read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</syntaxhighlight>


== [https://cran.r-project.org/web/packages/readr/ readr] (it is not designed to read Excel files) ==
=== [http://rdotnet.codeplex.com/ R.NET] ===


Compared to base equivalents like '''read.csv()''', '''readr''' is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.


[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released.
Terminal
 
{{Pre}}
The '''read_csv()''' function from the '''readr''' package is as fast as '''fread()''' function from '''data.table''' package. ''For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv().'' See 5.3 of Efficient R Programming book.
# jdk 7
 
sudo apt-get install openjdk-7-*
Note that '''fread()''' can read-n a selection of the columns.
update-alternatives --config java
 
# oracle jdk 8
== Colors ==
sudo add-apt-repository -y ppa:webupd8team/java
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
sudo apt-get update
* [http://colorspace.r-forge.r-project.org/articles/colorspace.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes.
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
* palette_explorer() function from the [https://cran.r-project.org/web/packages/tmaptools/index.html tmaptools] package. See [https://www.computerworld.com/article/3184778/data-analytics/6-useful-r-functions-you-might-not-know.html selecting color palettes with shiny].
sudo apt-get -y install openjdk-8-jdk
* [http://www.ucl.ac.uk/~zctpep9/Archived%20webpages/Cookbook%20for%20R%20%C2%BB%20Colors%20(ggplot2).htm Cookbook for R]
</pre>
* [http://ggplot2.tidyverse.org/reference/scale_brewer.html Sequential, diverging and qualitative colour scales/palettes from colorbrewer.org]: scale_colour_brewer(), scale_fill_brewer(), ...
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* http://colorbrewer2.org/
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
* It seems there is no choice of getting only 2 colors no matter which set name we can use
<pre>
* To see the set names used in brewer.pal, see
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
** RColorBrewer::display.brewer.all()
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
** For example, [http://colorbrewer2.org/#type=qualitative&scheme=Set1&n=4 Set1] from http://colorbrewer2.org/
</pre>
* To list all R color names, colors().
* And then run '''sudo ldconfig'''
** [http://research.stowers.org/mcm/efg/R/Color/Chart/ColorChart.pdf Color Chart] (include Hex and RGB) & [http://research.stowers.org/mcm/efg/Report/UsingColorInR.pdf Using Color in R] from http://research.stowers.org
** Code to generate rectangles with colored background https://www.r-graph-gallery.com/42-colors-names/
* [https://stackoverflow.com/questions/28461326/convert-hex-color-code-to-color-name convert hex value to color names] <syntaxhighlight lang='rsplus'>
library(plotrix)
sapply(rainbow(4), color.id)
sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
</syntaxhighlight>
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.3/topics/hsv hsv()] function. [https://eranraviv.com/matrix-style-screensaver-in-r/ Matrix-style screensaver in R]


Below is an example using the option ''scale_fill_brewer''(palette = "[http://colorbrewer2.org/#type=qualitative&scheme=Paired&n=9 Paired]"). See the source code at [https://gist.github.com/JohannesFriedrich/c7d80b4e47b3331681cab8e9e7a46e17 gist]. Note that only 'set1' and 'set3' palettes in '''qualitative scheme''' can support up to 12 classes.
Now go back to R
{{Pre}}
install.packages("rJava")
</pre>
Done!


According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
If above does not work, a simple way is by (under Ubuntu) running
<pre>
sudo apt-get install r-cran-rjava
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


[[File:GgplotPalette.svg|300px]]
=== RCaller ===


== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
=== RApache ===
See [[Ggplot2|ggplot2]]
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf


== Data Manipulation & Tidyverse ==
=== Rscript, arguments and commandArgs() ===
* [https://www.tidyverse.org/ Tidyverse] Homepage
[https://www.r-bloggers.com/passing-arguments-to-an-r-script-from-command-lines/ Passing arguments to an R script from command lines]
* [https://www.rstudio.com/resources/webinars/pipelines-for-data-analysis-in-r/ Pipelines for data analysis in R], [https://www.rstudio.com/resources/videos/data-science-in-the-tidyverse/ Data Science in the Tidyverse]
Syntax:
<pre>
<pre>
  Import
$ Rscript --help
    |
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
    | readr, readxl
    | haven, DBI, httr  +----- Visualize ------+
    |                    |    ggplot2, ggvis    |
    |                    |                      |
  Tidy ------------- Transform
  tibble              dplyr                  Model
  tidyr                  |                   broom
                          +------ Model ---------+
</pre>
</pre>
* [http://r4ds.had.co.nz/ R for Data Science] and [http://tidyverse.org/ tidyverse] package (it is a collection of '''ggplot2, tibble, tidyr, readr, purrr''' & '''dplyr''' packages).
** tidyverse, among others, was used at [http://juliasilge.com/blog/Mining-CRAN-DESCRIPTION/ Mining CRAN DESCRIPTION Files] (tbl_df(), %>%, summarise(), count(), mutate(), arrange(), unite(), ggplot(), filter(), select(), ...). Note that there is a problem to reproduce the result. I need to run ''cran <- cran[, -14]'' to remove the MD5sum column.
** [http://brettklamer.com/diversions/statistical/compile-r-for-data-science-to-a-pdf/ Compile R for Data Science to a PDF]
* [https://www.rstudio.com/wp-content/uploads/2015/02/data-wrangling-cheatsheet.pdf Data Wrangling with dplyr and tidyr Cheat Sheet]
* [https://hbctraining.github.io/Intro-to-R/lessons/07_intro_tidyverse.html Data Wrangling with Tidyverse] from the Harvard Chan School of Public Health.
* [http://datascienceplus.com/best-packages-for-data-manipulation-in-r/ Best packages for data manipulation in R]. It demonstrates to perform the same tasks using '''data.table''' and '''dplyr''' packages. '''data.table''' is faster and it may be a go-to package when performance and memory are the constraints.
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by Alboukadel Kassambara
** subset data frame columns: pull(), select(), select_if(), other helper functions
** subset (filter) data frame rows: slice(), filter(), filter_all(), filter_if(), filter_at(), sample_n(), top_n()
** identify and remove duplicate rows: duplicated(), unique(), distinct()
** ordering rows: arrange(), desc()
** renaming and adding columns: rename()
** compute and add new variables to a data frame: mutate(), transmutate()
** computing summary statistics (pay to view)


=== Install on Ubuntu ===
Example:
[https://stackoverflow.com/a/46983233 How to install Tidyverse on Ubuntu 16.04 and 17.04]
<pre>
<syntaxhighlight lang='bash'>
args = commandArgs(trailingOnly=TRUE)
# Ubuntu >= 18.04. However, I get unmet dependencies errors on R 3.5.3.
# test if there is at least one argument: if not, return an error
# r-cran-curl : Depends: r-api-3.4
if (length(args)==0) {
sudo apt-get install r-cran-curl r-cran-openssl r-cran-xml2
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
} else if (length(args)==1) {
  # default output file
  args[2] = "out.txt"
}
cat("args[1] = ", args[1], "\n")
cat("args[2] = ", args[2], "\n")
</pre>
<pre>
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt
# args[2] =  out.txt
</pre>


# Works fine on Ubuntu 16.04, 18.04
=== Rscript, #! Shebang and optparse package ===
sudo apt install libcurl4-openssl-dev libssl-dev libxml2-dev
<ul>
</syntaxhighlight>
<li>Writing [https://www.r-bloggers.com/2014/05/r-scripts/ R scripts] like linux bash files.
<li>[https://www.makeuseof.com/shebang-in-linux/ What Is the Shebang (#!) Character Sequence in Linux?]
<li>[https://blog.rmhogervorst.nl/blog/2020/04/14/where-does-the-output-of-rscript-go/ Where does the output of Rscript go?]
<li>Create a file <shebang.R>.
<pre>
#!/usr/bin/env Rscript
print ("shebang works")
</pre>
Then in the command line
<pre>
chmod u+x shebang.R
./shebang.R
</pre>
<li>[http://www.cureffi.org/2014/01/15/running-r-batch-mode-linux/ Running R in batch mode on Linux]
<li>[https://cran.r-project.org/web/packages/optparse/index.html optparse] package. Check out its vignette.
<li>[https://cran.r-project.org/web/packages/getopt/index.html getopt]: C-Like 'getopt' Behavior.
</ul>


80 R packages will be installed after ''tidyverse'' has been installed.
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
Provides hash-bang (#!) capability for R


=== Install on Raspberry Pi/(ARM based) Chromebook ===
FAQs:
In additional to the requirements of installing on Ubuntu, I got an error when it is installing a dependent package [https://github.com/r-lib/fs/issues/146 fs]: '''undefined symbol: pthread_atfork'''. The [https://cran.r-project.org/web/packages/fs/index.html fs] package version is 1.2.6. The [https://github.com/r-lib/fs/issues/128#issuecomment-435552967 solution] is to add one line in fs/src/Makevars file and then install the "fs" package using the source on the local machine.
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
{{Pre}}
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"


=== [http://rpubs.com/danmirman/Rgroup-part1 5 most useful data manipulation functions] ===
-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
* subset() for making subsets of data (natch)
                                              # Example: R -q -e "date()"
* merge() for combining data sets in a smart and easy way
* '''melt()'''-reshape2 package for converting from wide to long data formats. See an example [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others here] where we want to combine multiple columns of values into 1 column. melt() is replaced by gather().
* dcast()-reshape2 package for converting from long to wide data formats (or just use [https://datascienceplus.com/building-barplots-with-error-bars/ tapply()]), and for making summary tables
* ddply()-plyr package for doing split-apply-combine operations, which covers a huge swath of the most tricky data operations


=== [https://cran.r-project.org/web/packages/data.table/index.html data.table] ===
-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
Fast aggregation of large data (e.g. 100GB in RAM or just several GB size file), fast ordered joins, fast add/modify/delete of columns by group using no copies at all, list columns and a fast file reader (fread).
                                              # It won't show the startup message when it is used in the command line.
                                              # Example: Rscript -e "date()"
</pre>


Some resources:
We can install littler using two ways.
* https://www.rdocumentation.org/packages/data.table/versions/1.12.0
* install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ''~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r''). You need to create a soft link in order to access it globally.
* [https://www.waldrn.com/dplyr-vs-data-table/ R Packages: dplyr vs data.table]
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.
* [https://github.com/rstudio/cheatsheets/raw/master/datatable.pdf Cheat sheet] from [https://www.rstudio.com/resources/cheatsheets/ RStudio]
* [https://www.r-bloggers.com/importing-data-into-r-part-two/ Reading large data tables in R]. fread(FILENAME)
* Note that '''x[, 2]'' always return 2. If you want to do the thing you want, use ''x[, 2, with=FALSE]'' or ''x[, V2]'' where V2 is the header name. See the FAQ #1 in [http://datatable.r-forge.r-project.org/datatable-faq.pdf data.table].
* [http://r-norberg.blogspot.com/2016/06/understanding-datatable-rolling-joins.html Understanding data.table Rolling Joins]
* [https://rollingyours.wordpress.com/2016/06/14/fast-aggregation-of-large-data-with-the-data-table-package/ Intro to The data.table Package]
** Subsetting rows and/or columns
** Alternative to using tapply(), aggregate(), table() to summarize data
** Similarities to SQL, DT[i, j, by]
* [https://www.listendata.com/2016/10/r-data-table.html R : data.table (with 50 examples)] from ListenData
** Describe Data
** Selecting or Keeping Columns
** Rename Variables
** Subsetting Rows / Filtering
** Faster Data Manipulation with Indexing
** Performance Comparison
** Sorting Data
** Adding Columns (Calculation on rows)
** How to write Sub Queries (like SQL)
** Summarize or Aggregate Columns
** GROUP BY (Within Group Calculation)
** Remove Duplicates
** Extract values within a group
** SQL's RANK OVER PARTITION
** Cumulative SUM by GROUP
** Lag and Lead
** Between and LIKE Operator
** Merging / Joins
** Convert a data.table to data.frame
* [https://www.dezyre.com/data-science-in-r-programming-tutorial/r-data-table-tutorial R Tutorial: data.table] from dezyre.com
** Syntax: DT[where, select|update|do, by]
** Keys and setkey()
** Fast grouping using j and by: DT[,sum(v),by=x]
** Fast ordered joins: X[Y,roll=TRUE]
* In the [https://cran.r-project.org/web/packages/data.table/vignettes/datatable-intro-vignette.html Introduction to data.table] vignette, the data.table::order() function is SLOWER than base::order() from my Odroid xu4 (running Ubuntu 14.04.4 trusty on uSD) <syntaxhighlight lang='rsplus'>
odt = data.table(col=sample(1e7))
(t1 <- system.time(ans1 <- odt[base::order(col)]))  ## uses order from base R
#  user  system elapsed
#  2.730  0.210  2.947
(t2 <- system.time(ans2 <- odt[order(col)]))        ## uses data.table's order
#  user  system elapsed
#  2.830  0.215  3.052
(identical(ans1, ans2))
# [1] TRUE
</syntaxhighlight>
* [https://jangorecki.github.io/blog/2016-06-30/Boost-Your-Data-Munging-with-R.html Boost Your Data Munging with R]
* [https://www.rdocumentation.org/packages/data.table/versions/1.12.0/topics/rbindlist rbindlist()]. One problem, it uses too much memory. In fact, when I try to analyze R package downloads, the command "dat <- rbindlist(logs)" uses up my 64GB memory (OS becomes unresponsive).


[https://github.com/Rdatatable/data.table/wiki/Installation#openmp-enabled-compiler-for-mac OpenMP enabled compiler for Mac]. This instruction works on my Mac El Capitan (10.11.6) when I need to upgrade the data.table version from 1.11.4 to 1.11.6.
After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The [http://dirk.eddelbuettel.com/code/littler.examples.html web version of examples] does not have the TOC.


Question: how to make use multicore with data.table package?
'''r''' was not meant to run interactively like '''R'''. See ''man r''.


=== reshape & reshape2 (superceded by tidyr package) ===
=== RInside: Embed R in C++ ===
* [http://r-exercises.com/2016/07/06/data-shape-transformation-with-reshape/ Data Shape Transformation With Reshape()]
See [[R#RInside|RInside]]
* Use '''acast()''' function in reshape2 package. It will convert data.frame used for analysis to a table-like data.frame good for display.
* http://lamages.blogspot.com/2013/10/creating-matrix-from-long-dataframe.html


=== [http://cran.r-project.org/web/packages/tidyr/index.html tidyr] and benchmark ===
(''From RInside documentation'') The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.
An evolution of reshape2. It's designed specifically for data tidying (not general reshaping or aggregating) and works well with dplyr data pipelines.


* [https://cran.r-project.org/web/packages/tidyr/vignettes/tidy-data.html vignette("tidy-data")] & [https://github.com/rstudio/cheatsheets/blob/master/data-import.pdf Cheat sheet]
The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
* Main functions
** Reshape data: [https://www.rdocumentation.org/packages/tidyr/versions/0.8.3/topics/gather gather()] & [https://www.rdocumentation.org/packages/tidyr/versions/0.8.3/topics/spread spread()]. [https://tidyr.tidyverse.org/dev/articles/pivot.html These two will be deprecated]
** Split cells: [https://www.rdocumentation.org/packages/tidyr/versions/0.8.3/topics/separate separate()] & [https://www.rdocumentation.org/packages/tidyr/versions/0.8.3/topics/unite unite()]
** Handle missing: drop_na() & fill() & replace_na()
* Other functions
** tidyr::separate() function. If a cell contains many elements separated by ",", we can use this function to create more columns. The opposite function is unite().
** [https://rpubs.com/maraaverick/tidyr-separate-rows-FTW tidyr::separate_rows()]. If a cell contains many elements separated by ",", we can use this function to create one more row. See the cheat sheet link above.
* http://blog.rstudio.org/2014/07/22/introducing-tidyr/
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
* http://timelyportfolio.github.io/rCharts_factor_analytics/factors_with_new_R.html
* [http://www.milanor.net/blog/reshape-data-r-tidyr-vs-reshape2/ tidyr vs reshape2]
* [https://data.library.virginia.edu/a-tidyr-tutorial/ A tidyr Tutorial] from U of Virginia
* [http://r-posts.com/benchmarking-cast-in-r-from-long-data-frame-to-wide-matrix/ Benchmarking cast in R from long data frame to wide matrix]


Make wide tables long with '''gather()''' (see 6.3.1 of Efficient R Programming)
To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.
<syntaxhighlight lang='rsplus'>
library(tidyr)
library(efficient)
data(pew) # wide table
dim(pew) # 18 x 10, (religion, '<$10k', '$10--20k', '$20--30k', ..., '>150k')
pewt <- gather(data = pew, key = Income, value = Count, -religion)
dim(pew) # 162 x 3,  (religion, Income, Count)


args(gather)
To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
# function(data, key, value, ..., na.rm = FALSE, convert = FALSE, factor_key = FALSE)
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>
</syntaxhighlight>
where the three arguments of gather() requires:
* data: a data frame in which column names will become row vaues
* key: the name of the categorical variable into which the column names in the original datasets are converted.
* value: the name of cell value columns


In this example, the 'religion' column will not be included (-religion).
The real build process looks like (check <Makefile> for completeness)
<pre>
g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include  \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0
</pre>


=== dplyr, plyr packages ===
Hello World example of embedding R in C++.
* plyr package suffered from being slow in some cases. dplyr addresses this by porting much of the computation to C++. Another additional feature is the ability to work with data stored directly in an external '''database'''. The benefits of doing this are the data can be managed natively in a relational database, queries can be conducted on that database, and only the results of query returned.
<pre>
* Essential functions: 3 rows functions, 3 column functions and 1 mixed function.
#include <RInside.h>                   // for the embedded R via RInside
: <syntaxhighlight lang='rsplus'>
          select, mutate, rename
            +------------------+
filter      +                  +
arrange    +                  +
group_by    +                  +
drop_na    +                  +
ungroup    + summarise        +
            +------------------+
</syntaxhighlight>
* These functions works on data frames and tibble objects.
: <syntaxhighlight lang='rsplus'>
iris %>% filter(Species == "setosa") %>% count()
head(iris %>% filter(Species == "setosa") %>% arrange(Sepal.Length))
</syntaxhighlight>
* [http://genomicsclass.github.io/book/pages/dplyr_tutorial.html dplyr tutorial] from PH525x series (Biomedical Data Science by Rafael Irizarry and Michael Love). For select() function, some additional options to select columns based on a specific criteria include
** start_with()/ ends_with() = Select columns that start/end with a character string
** contains() = Select columns that contain a character string
** matches() = Select columns that match a regular expression
** one_of() = Select columns names that are from a group of names
* [http://r4ds.had.co.nz/transform.html Data Transformation] in the book '''R for Data Science'''. Five key functions in the '''dplyr''' package:
** Filter rows: filter()
** Arrange rows: arrange()
** Select columns: select()
** Add new variables: mutate()
** Grouped summaries: group_by() & summarise()
: <syntaxhighlight lang='rsplus'>
# filter
jan1 <- filter(flights, month == 1, day == 1)
filter(flights, month == 11 | month == 12)
filter(flights, arr_delay <= 120, dep_delay <= 120)
df <- tibble(x = c(1, NA, 3))
filter(df, x > 1)
filter(df, is.na(x) | x > 1)


# arrange
int main(int argc, char *argv[]) {
arrange(flights, year, month, day)
arrange(flights, desc(arr_delay))


# select
    RInside R(argc, argv);              // create an embedded R instance
select(flights, year, month, day)
select(flights, year:day)
select(flights, -(year:day))


# mutate
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
flights_sml <- select(flights,
  year:day,
  ends_with("delay"),
  distance,
  air_time
)
mutate(flights_sml,
  gain = arr_delay - dep_delay,
  speed = distance / air_time * 60
)
# if you only want to keep the new variables
transmute(flights,
  gain = arr_delay - dep_delay,
  hours = air_time / 60,
  gain_per_hour = gain / hours
)


# summarise()
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
by_day <- group_by(flights, year, month, day)
summarise(by_day, delay = mean(dep_delay, na.rm = TRUE))


# pipe. Note summarise() can return more than 1 variable.
     exit(0);
delays <- flights %>%
}
  group_by(dest) %>%
</pre>
  summarise(
     count = n(),
    dist = mean(distance, na.rm = TRUE),
    delay = mean(arr_delay, na.rm = TRUE)
  ) %>%
  filter(count > 20, dest != "HNL")
flights %>%
  group_by(year, month, day) %>%
  summarise(mean = mean(dep_delay, na.rm = TRUE))
</syntaxhighlight>
* Videos
** [https://youtu.be/jWjqLW-u3hc Hands-on dplyr tutorial for faster data manipulation in R] by Data School. At time 17:00, it compares the '''%>%''' operator, '''with()''' and '''aggregate()''' for finding group mean.
** https://youtu.be/aywFompr1F4 (shorter video) by Roger Peng
** https://youtu.be/8SGif63VW6E by Hadley Wickham
** [https://www.rstudio.com/resources/videos/tidy-eval-programming-with-dplyr-tidyr-and-ggplot2/ Tidy eval: Programming with dplyr, tidyr, and ggplot2]. Bang bang "!!" operator was introduced for use in a function call.
* [https://csgillespie.github.io/efficientR/data-carpentry.html#dplyr Efficient R Programming]
* [http://www.r-exercises.com/2017/07/19/data-wrangling-transforming-23/ Data wrangling: Transformation] from R-exercises.
* [https://rollingyours.wordpress.com/2016/06/29/express-intro-to-dplyr/ Express Intro to dplyr] by rollingyours.
* [https://martinsbioblogg.wordpress.com/2017/05/21/using-r-when-using-do-in-dplyr-dont-forget-the-dot/ the dot].
* [http://martinsbioblogg.wordpress.com/2013/03/24/using-r-reading-tables-that-need-a-little-cleaning/ stringr and plyr] A '''data.frame''' is pretty much a list of vectors, so we use plyr to apply over the list and stringr to search and replace in the vectors.
* https://randomjohn.github.io/r-maps-with-census-data/ dplyr and stringr are used
* [https://datascienceplus.com/5-interesting-subtle-insights-from-ted-videos-data-analysis-in-r/ 5 interesting subtle insights from TED videos data analysis in R]
* [https://www.mango-solutions.com/blog/what-is-tidy-eval-and-why-should-i-care What is tidy eval and why should I care?]


=== stringr ===
The above can be compared to the Hello world example in Qt.
* https://www.rstudio.com/wp-content/uploads/2016/09/RegExCheatsheet.pdf
<pre>
* [https://github.com/rstudio/cheatsheets/raw/master/strings.pdf stringr Cheat sheet] (2 pages, this will immediately download the pdf file)
#include <QApplication.h>
* [https://csgillespie.github.io/efficientR/data-carpentry.html#regular-expressions Efficient data carpentry &#8594; Regular expressions] from Efficient R programming book by Gillespie & Lovelace.
#include <QPushButton.h>


=== [https://github.com/smbache/magrittr magrittr] ===
int main( int argc, char **argv )
* [https://cran.r-project.org/web/packages/magrittr/vignettes/magrittr.html Vignettes]
{
* [http://www.win-vector.com/blog/2018/04/magrittr-and-wrapr-pipes-in-r-an-examination/ magrittr and wrapr Pipes in R, an Examination]
    QApplication app( argc, argv );


Instead of nested statements, it is using pipe operator '''%>%'''. So the code is easier to read. Impressive!
     QPushButton hello( "Hello world!", 0 );
<syntaxhighlight lang='rsplus'>
    hello.resize( 100, 30 );
x %>% f     # f(x)
x %>% f(y)  # f(x, y)
x %>% f(arg=y)  # f(x, arg=y)
x %>% f(z, .) # f(z, x)
x %>% f(y) %>% g(z)  #  g(f(x, y), z)


x %>% select(which(colSums(!is.na(.))>0))  # remove columns with all missing data
    app.setMainWidget( &hello );
x %>% select(which(colSums(!is.na(.))>0)) %>% filter((rowSums(!is.na(.))>0)) # remove all-NA columns _and_ rows
    hello.show();
</syntaxhighlight>
* [http://www.win-vector.com/blog/2018/03/r-tip-make-arguments-explicit-in-magrittr-dplyr-pipelines/ Make Arguments Explicit in magrittr/dplyr Pipelines]
<syntaxhighlight lang='rsplus'>
suppressPackageStartupMessages(library("dplyr"))
starwars %>%
  filter(., height > 200) %>%
  select(., height, mass) %>%
  head(.)
# instead of
starwars %>%
  filter(height > 200) %>%
  select(height, mass) %>%
  head
</syntaxhighlight>
* [https://stackoverflow.com/questions/27100678/how-to-extract-subset-an-element-from-a-list-with-the-magrittr-pipe Subset an element from a list]
<syntaxhighlight lang='rsplus'>
iris$Species
iris[["Species"]]


iris %>%
    return app.exec();
`[[`("Species")
}
</pre>


iris %>%
=== [http://www.rfortran.org/ RFortran] ===
`[[`(5)
RFortran is an open source project with the following aim:


iris %>%
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
  subset(select = "Species")
</syntaxhighlight>
* '''Split-apply-combine''': group + summarize + sort/arrange + top n. The following example is from [https://csgillespie.github.io/efficientR/data-carpentry.html#data-aggregation Efficient R programming].
<syntaxhighlight lang='rsplus'>
data(wb_ineq, package = "efficient")
wb_ineq %>%
  filter(grepl("g", Country)) %>%
  group_by(Year) %>%
  summarise(gini = mean(gini, na.rm  = TRUE)) %>%
  arrange(desc(gini)) %>%
  top_n(n = 5)
</syntaxhighlight>
* [https://drdoane.com/writing-pipe-friendly-functions/ Writing Pipe-friendly Functions]
* http://rud.is/b/2015/02/04/a-step-to-the-right-in-r-assignments/
* http://rpubs.com/tjmahr/pipelines_2015
* http://danielmarcelino.com/i-loved-this-crosstable/
* http://moderndata.plot.ly/using-the-pipe-operator-in-r-with-plotly/
* Videos
** [https://www.rstudio.com/resources/videos/writing-readable-code-with-pipes/ Writing Readable Code with Pipes]
** [https://youtu.be/iIBTI_qiq9g Pipes in R - An Introduction to magrittr package]
<syntaxhighlight lang='rsplus'>
# Examples from R for Data Science-Import, Tidy, Transform, Visualize, and Model
diamonds <- ggplot2::diamonds
diamonds2 <- diamonds %>% dplyr::mutate(price_per_carat = price / carat)


pryr::object_size(diamonds)
It works only on Windows platform with Microsoft Visual Studio installed:(
pryr::object_size(diamonds2)
pryr::object_size(diamonds, diamonds2)


rnorm(100) %>% matrix(ncol = 2) %>% plot() %>% str()
== Call R from other languages ==
rnorm(100) %>% matrix(ncol = 2) %T>% plot() %>% str() # 'tee' pipe
=== C ===
    # %T>% works like %>% except that it returns the lefthand side (rnorm(100) %>% matrix(ncol = 2)) 
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]
    # instead of the righthand side.


# If a function does not have a data frame based api, you can use %$%.
Error: [https://stackoverflow.com/questions/43662542/not-resolved-from-current-namespace-error-when-calling-c-routines-from-r “not resolved from current namespace” error, when calling C routines from R]
# It explodes out the variables in a data frame.
mtcars %$% cor(disp, mpg)


# For assignment, magrittr provides the %<>% operator
Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
mtcars <- mtcars %>% transform(cyl = cyl * 2) # can be simplified by
mtcars %<>% transform(cyl = cyl * 2)
</syntaxhighlight>


Upsides of using magrittr: no need to create intermediate objects, code is easy to read.
=== JRI ===
http://www.rforge.net/JRI/


When not to use the pipe
=== ryp2 ===
* your pipes are longer than (say) 10 steps
http://rpy.sourceforge.net/rpy2.html
* you have multiple inputs or outputs
* Functions that use the current environment: assign(), get(), load()
* Functions that use lazy evaluation: tryCatch(), try()


=== outer() ===
== Create a standalone Rmath library ==
R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of [http://cran.r-project.org/doc/manuals/R-admin.html#The-standalone-Rmath-library R-admin manual].


=== Genomic sequence ===
Here is my experience based on R 3.0.2 on Windows OS.
* chartr
<syntaxhighlight lang='bash'>
> yourSeq <- "AAAACCCGGGTTTNNN"
> chartr("ACGT", "TGCA", yourSeq)
[1] "TTTTGGGCCCAAANNN"
</syntaxhighlight>


=== broom ===
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
[https://cran.r-project.org/web/packages/broom/index.html broom]: Convert Statistical Analysis Objects into Tidy Tibbles
Suppose we have downloaded R source code and build R from its source. See [[R#Build_R_from_its_source|Build_R_from_its_source]]. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


=== lobstr package - dig into the internal representation and structure of R objects ===
=== Use Rmath library in our code ===
[https://www.tidyverse.org/articles/2018/12/lobstr/ lobstr 1.0.0]
<pre>
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.


== Data Science ==
# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
See [[Data_science|Data science]] page
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# It is OK to save the cpp file under any directory.


== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ==
# Force to link against the static library <libRmath.a>
If we want to create the image on this wiki left hand side panel, we can use the '''jpeg''' package to read an existing plot and then edit and save it.
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


We can also use the jpeg package to import and manipulate a jpg image. See [http://moderndata.plot.ly/fun-with-heatmaps-and-plotly/ Fun with Heatmaps and Plotly].
# Force to link against dynamic library <Rmath.dll>
 
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ==
</pre>
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
 
== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ==
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
 
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
<pre>
<pre>
sudo apt-get install libgtk2.0-dev libcairo2-dev
c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>
</pre>


On Windows OS, we may got the error: '''unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' '''. We need to follow the instruction in [http://tolstoy.newcastle.edu.au/R/e6/help/09/05/15613.html here].
Below is the cpp program <RmathEx1.cpp>.
<pre>
//RmathEx1.cpp
#define MATHLIB_STANDALONE
#include <iostream>
#include "Rmath.h"


== [http://igraph.org/r/ igraph] ==
using std::cout; using std::cin; using std::endl;
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]


== Identifying dependencies of R functions and scripts ==
int main()
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
{
<syntaxhighlight lang='rsplus'>
  double x1, x2;
library(mvbutils)
  cout << "Enter a argument for the normal cdf:" << endl;
foodweb(where = "package:batr")
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
  cout << "Prob(Z <= " << x1 << ") = " <<
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
</pre>


foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
== Calling R.dll directly ==
</syntaxhighlight>
See Chapter 8.2.2 of [http://cran.r-project.org/doc/manuals/R-exts.html#Calling-R_002edll-directly|Writing R Extensions]. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.


== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
== Create HTML report ==
Iterator is useful over for-loop if the data is already a '''collection'''. It can be used to iterate over a vector, data frame, matrix, file
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].


Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.


== Colors ==
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
* http://deanattali.com/blog/colourpicker-package/ Not sure what it is doing
* [http://gforge.se/2020/07/news-in-htmltable-2-0/ News in htmlTable 2.0]
* [http://www.lifehack.org/484519/how-to-choose-the-best-colors-for-your-data-charts How to Choose the Best Colors For Your Data Charts]
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* [http://sape.inf.usi.ch/quick-reference/ggplot2/colour Color names in R]


=== [http://rpubs.com/gaston/colortools colortools] ===
=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
Tools that allow users generate color schemes and palettes
* https://github.com/renkun-ken/formattable
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]


=== [https://github.com/daattali/colourpicker colourpicker] ===
=== [https://github.com/crubba/htmltab htmltab] package ===
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
This package is NOT used to CREATE html report but EXTRACT html table.


=== [https://cran.r-project.org/web/packages/inlmisc/index.html inlmisc] ===
=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
[https://owi.usgs.gov/blog/tolcolors/ GetTolColors()]. Lots of examples.
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.


== [https://github.com/kevinushey/rex rex] ==
== Create academic report ==
Friendly Regular Expressions
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.


== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
== Create pdf and epub files ==
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
{{Pre}}
# Idea:
#        knitr        pdflatex
#  rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
</pre>
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
{{Pre}}
git clone https://github.com/yihui/knitr-examples.git
</pre>
* <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run '''sudo apt-get install texlive-fonts-recommended''' to install missing fonts. It works!
 
To see a real example, check out [http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html DESeq2] package (inst/doc subdirectory). In addition to DESeq2, I also need to install '''DESeq, BiocStyle, airway, vsn, gplots''', and '''pasilla''' packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.


See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
{{Pre}}
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf


<pre>
git clone https://github.com/yihui/knitr-examples.git
library(formatR)
cd knitr-examples
tidy_source("Input.R", file = "output.R", width.cutoff=70)
R -e "library(knitr); knit('001-minimal.Rmd')"
tidy_source("clipboard")
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
# default width is getOption("width") which is 127 in my case.
</pre>
</pre>


Some issues
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
{{Pre}}
<pre>
# Idea:
cat("abcd",
#        knitr        pandoc
    # This is my comment
#  rnw -------> tex ----------> markdown or epub
    "defg")
 
library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</pre>
</pre>
will result in
 
Convert tex to epub
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
 
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
 
* https://rmarkdown.rstudio.com/lesson-7.html
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
* https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/
* [https://cran.r-project.org/web/packages/kableExtra/vignettes/awesome_table_in_html.html kableExtra] package
 
== Create Word report ==
 
=== Using the power of Word ===
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]
 
=== knitr + pandoc ===
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* http://rmarkdown.rstudio.com/articles_docx.html
 
It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.
<pre>
<pre>
> tidy_source("clipboard")
# Idea:
Error in base::parse(text = code, srcfile = NULL) :
#        knitr      pandoc
  3:1: unexpected string constant
#  rmd -------> md --------> docx
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
library(knitr)
3: "defg"
knit2html("example.rmd") #Create md and html files
  ^
</pre>
</pre>
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
and then
<pre>
<pre>
cat("abcd"
FILE <- "example"
    ,"defg"   # This is my comment
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
  ,"ghij")
</pre>
</pre>
will become
Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.
 
Another way is
<pre>
<pre>
cat("abcd", "defg" # This is my comment
library(pander)
, "ghij")  
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
</pre>
</pre>
Still bad!!
 
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:
* A pdf file: pandoc -s report.md -t latex -o report.pdf
* A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
* Openoffice: pandoc report.md -o report.odt
* Word docx: pandoc report.md -o report.docx
 
We can also create the epub file for reading on Kobo ereader. For example, download [https://gist.github.com/jeromyanglim/2716336 this file] and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!
<pre>
<pre>
cat("</p>",
knit("example.Rmd")
"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
pandoc("example.md", format="epub")
ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),
file=ExternalFileName, sep="\n", append=T)
</pre>
</pre>
will result in
 
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<pre>
<pre>
> tidy_source("clipboard", width.cutoff=70)
> pandoc("Rmd_to_Epub.md", format="epub")
Error in base::parse(text = code, srcfile = NULL) :  
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
  3:129: unexpected SPECIAL
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
Error in (function (input, format, ext, cfg) : conversion failed
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
In addition: Warning message:
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
</pre>
</pre>
* ''width.cutoff'' parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
 
=== pander ===
Try pandoc[1] with a minimal reproducible example, you might give a try to my "[http://cran.r-project.org/web/packages/pander/ pander]" package [2] too:
 
<pre>
<pre>
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",  
library(pander)
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,  
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
    sep = "\n", append = T)
            output = tempfile(), convert = 'docx')
</pre>
</pre>
* It merges lines though I don't always want to do that. For example
Where the content of the "minimal.brew" file is something you might have
got used to with Sweave - although it's using "brew" syntax instead. See
the examples of pander [3] for more details. Please note that pandoc should
be installed first, which is pretty easy on Windows.
 
# http://johnmacfarlane.net/pandoc/
# http://rapporter.github.com/pander/
# http://rapporter.github.com/pander/#examples
 
=== R2wd ===
Use [http://cran.r-project.org/web/packages/R2wd/ R2wd] package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.  
<pre>
<pre>
cat("abcd"
> library(R2wd)
    ,"defg" 
> wdGet()
  ,"ghij")
Loading required package: rcom
</pre>
Loading required package: rscproxy
will become
rcom requires a current version of statconnDCOM installed.
<pre>
To install statconnDCOM type
cat("abcd", "defg", "ghij")  
    installstatconnDCOM()
 
This will download and install the current version of statconnDCOM
 
You will need a working Internet connection
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
  argument is of length zero
</pre>
</pre>


== Download papers ==
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
Search and Download Papers from the bioRxiv Preprint Server


=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
=== Convert from pdf to word ===
Interface to the arXiv API
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert


=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
=== rtf ===
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
* https://ropensci.org/technotes/2018/12/14/pdftools-20/


== [https://github.com/ColinFay/aside aside]: set it aside ==
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
An RStudio addin to run long R commands aside your current session.
Package xtable will produce html output.  
{{Pre}}
print(xtable(X), type="html")
</pre>


== Teaching ==
If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics


== Organize R research project ==
=== officer ===
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
<ul>
* [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow]. It suggests the [https://github.com/r-lib/here here] package. Don't use '''setwd()''' and '''rm(list = ls())'''.
<li>[https://cran.r-project.org/web/packages/officer/index.html CRAN]. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.  
** In RStudio, if we try to send a few lines of code and one of the line contains '''setwd()''', it will give a message: ''The working directory was changed to XXX inside a notebook chunk. The working directory will be reset when the chunk is finished running. Use the knitr root.dir option in the setup chunk to change the working directory for notebook chunks.''
<li>The [https://gist.github.com/arraytools/4f182b036ae7f95a31924ba5d5d3f069 gist] includes a comprehensive example that encompasses various elements such as sections, subsections, and tables. It also incorporates a detailed paragraph, along with visual representations created using base R plots and ggplots.  
** No update for the ''here'' package. Consider [https://github.com/r-lib/usethis usethis] package.
<li>Add a line space
* drake project
<pre>
** [https://ropensci.org/blog/2018/02/06/drake/ The prequel to the drake R package]
doc <- body_add_par(doc, "")
** [https://ropenscilabs.github.io/drake-manual/index.html The drake R Package User Manual]


=== How to save (and load) datasets in R (.Rdata vs .Rds file) ===
# Function to add n line spaces
[https://rcrastinate.rbind.io/post/how-to-save-and-load-data-in-r-an-overview/ How to save (and load) datasets in R: An overview]
body_add_par_n <- function (doc, n) {
  for(i in 1:n){
    doc <- body_add_par(doc, "")
  }
  return(doc)
}
body_add_par_n(3)
</pre>
<li>[https://ardata-fr.github.io/officeverse/officer-for-word.html Figures] from the documentation of '''officeverse'''.
<li>See [https://stackoverflow.com/a/25427314 Data frame to word table?].
<li>See [[Office#Tables|Office]] page for some code.
<li>[https://www.r-bloggers.com/2020/07/how-to-read-and-create-word-documents-in-r/ How to read and create Word Documents in R] where we can extracting tables from Word Documents.
<pre>
x = read_docx("myfile.docx")
content <- docx_summary(x) # a vector
grep("nlme", content$text, ignore.case = T, value = T)
</pre>
</ul>


== Text to speech ==
== Powerpoint ==
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]
<ul>
<li>[https://cran.r-project.org/web/packages/officer/index.html officer] package  (formerly ReporteRs). [http://theautomatic.net/2020/07/28/how-to-create-powerpoint-reports-with-r/ How to create powerpoint reports with R]
</li>
<li>[https://davidgohel.github.io/flextable/ flextable] (imports '''officer''')
</li>
<li>[https://stackoverflow.com/a/21558466 R data.frame to table image for presentation].
<pre>
library(gridExtra)
grid.newpage()
grid.table(mydf)
</pre>
</li>
<li>[https://bookdown.org/yihui/rmarkdown/powerpoint-presentation.html Rmarkdown]
</li>
</ul>


== Weather data ==
== PDF manipulation ==
* [https://github.com/ropensci/prism prism] package
[https://github.com/pridiltal/staplr staplr]
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== logR ==
== R Graphs Gallery ==
https://github.com/jangorecki/logR
* [https://www.facebook.com/pages/R-Graph-Gallery/169231589826661 Romain François]
* [http://shinyapps.stat.ubc.ca/r-graph-catalog/ R Graph Catalog] written using R + Shiny. The source code is available on [https://github.com/jennybc/r-graph-catalog Github].
* Forest plot. See the packages [https://cran.r-project.org/web/packages/rmeta/index.html rmeta] and [https://cran.r-project.org/web/packages/forestplot/ forestplot]. The forest plot can be used to plot the quantities like relative risk (with 95% CI) in survival data.
** [http://www.danieldsjoberg.com/bstfun/dev/reference/add_inline_forest_plot.html Inline forest plot]


== Progress bar ==
== COM client or server ==
https://github.com/r-lib/progress#readme


Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.
=== Client ===


== cron ==
[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
[https://github.com/bnosac/cronr cronR]


= Different ways of using R =
=== Server ===
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


== 10 things R can do that might surprise you ==
== Use R under proxy ==
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


== R call C/C++ ==
== RStudio ==
Mainly talks about .C() and .Call().
* [https://github.com/rstudio/rstudio Github]
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]


Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.
=== rstudio.cloud ===
https://rstudio.cloud/


* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
=== Launch RStudio ===
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
* http://www.stat.harvard.edu/ccr2005/
* http://mazamascience.com/WorkingWithData/?p=1099


=== gcc ===
=== Create .Rproj file ===
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
If you have an existing package that doesn't have an .Rproj file, you can use '''devtools::use_rstudio("path/to/package")''' to add it.


== SEXP ==
With an RStudio project file, you can
Some examples from packages
* Restore .RData into workspace at startup
* Save workspace to .RData on exit (or '''save.image'''("Robj.RData") & load("Robj.RData"))
* Always save history (even if no saving .RData, '''savehistory'''(".Rhistory") & loadhistory(".Rhistory"))
* etc


* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
=== package search ===
https://github.com/RhoInc/CRANsearcher


== R call Fortran ==
=== Git ===
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
* [https://www.r-bloggers.com/the-need-for-speed-part-1-building-an-r-package-with-fortran-or-c/ The Need for Speed Part 1: Building an R Package with Fortran (or C)] (Very detailed)
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]
 
== Visual Studio ==
[http://blog.revolutionanalytics.com/2017/05/r-and-python-support-now-built-in-to-visual-studio-2017.html R and Python support now built in to Visual Studio 2017]


== Embedding R ==
== List files using regular expression ==
* Extension
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
list.files(pattern = "^Something")
</pre>


* See [http://cran.r-project.org/doc/manuals/R-exts.html#Linking-GUIs-and-other-front_002dends-to-R Writing for R Extensions] Manual Chapter 8.
Using '''Sys.glob()"' as
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
<pre>
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report]  by Friedrich Leisch in 2007.
> Sys.glob("~/Downloads/*.txt")
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>


=== An very simple example (do not return from shell) from Writing R Extensions manual ===
== Hidden tool: rsync in Rtools ==
The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe


This example can be run by
sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
total size is 1198416  speedup is 3.71


Note:  
c:\Rtools\bin>
# '''R_HOME/bin/exec/R''' is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
</pre>
# '''R_HOME/bin/R''' is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to ''/usr/local/bin/R''. When we run ''R_HOME/bin/R'', it actually runs ''R_HOME/bin/R CMD R_HOME/bin/exec/R'' (see line 259 of ''R_HOME/bin/R'' as in R 3.0.2) so we know the important role of ''R_HOME/bin/exec/R''.


More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also [https://superuser.com/a/69764 rsync file permissions on windows].


=== An example from Bioconductor workshop ===
== Install rgdal package (geospatial Data) on ubuntu ==
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
Terminal
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
{{Pre}}
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
</pre>
 
R
{{Pre}}
install.packages("rgdal")
</pre>


Example:
== Install sf package ==
Create <embed.c> file
I got the following error even I have installed some libraries.  
<pre>
<pre>
#include <Rembedded.h>
checking GDAL version >= 2.0.1... no
#include <Rdefines.h>
configure: error: sf is not compatible with GDAL versions below 2.0.1
</pre>
Then I follow the instruction here
{{Pre}}
sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable
 
sudo apt update
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears


static void doSplinesExample();
sudo apt install libgdal-dev # works on ubuntu 20.04 too
int
                            # no need the previous lines
main(int argc, char *argv[])
</pre>
{
    Rf_initEmbeddedR(argc, argv);
    doSplinesExample();
    Rf_endEmbeddedR(0);
    return 0;
}
static void
doSplinesExample()
{
    SEXP e, result;
    int errorOccurred;


    // create and evaluate 'library(splines)'
== Database ==
    PROTECT(e = lang2(install("library"), mkString("splines")));
* https://cran.r-project.org/web/views/Databases.html
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]
    if (errorOccurred) {
        // handle error
    }
    UNPROTECT(1);


    // 'options(FALSE)' ...
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
    PROTECT(e = lang2(install("options"), ScalarLogical(0)));
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
    // ... modified to 'options(example.ask=FALSE)' (this is obscure)
* https://github.com/rstats-db/RSQLite
    SET_TAG(CDR(e), install("example.ask"));
    R_tryEval(e, R_GlobalEnv, NULL);
    UNPROTECT(1);


    // 'example("ns")'
'''Creating a new database''':
    PROTECT(e = lang2(install("example"), mkString("ns")));
{{Pre}}
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
library(DBI)
    UNPROTECT(1);
 
}
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
</pre>
dbDisconnect(mydb)
Then build the executable. Note that I don't need to create R_HOME variable.
unlink("my-db.sqlite")
<pre>
cd
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest


nano embed.c
# temporary database
# Using a single line will give an error and cannot not show the real problem.
mydb <- dbConnect(RSQLite::SQLite(), "")
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
dbDisconnect(mydb)
# A better way is to run compile and link separately
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed
</pre>
</pre>


Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
'''Loading data''':
<pre>
{{Pre}}
export R_HOME=/home/brb/Downloads/R-3.0.2
mydb <- dbConnect(RSQLite::SQLite(), "")
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
dbWriteTable(mydb, "mtcars", mtcars)
./embed # No need to include R CMD in front.
dbWriteTable(mydb, "iris", iris)
 
dbListTables(mydb)
 
dbListFields(con, "mtcars")
 
dbReadTable(con, "mtcars")
</pre>
</pre>


Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].
'''Queries''':
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')


=== Create a Simple Socket Server in R ===
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].  


Create an R function
res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
<pre>
dbFetch(res)
simpleServer <- function(port=6543)
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}
</pre>
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.


Welcome to R!
'''Batched queries''':
R> summary(iris[, 3:5])
{{Pre}}
  Petal.Length    Petal.Width          Species 
dbClearResult(rs)
Min.   :1.000  Min.  :0.100  setosa    :50 
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
1st Qu.:1.600   1st Qu.:0.300  versicolor:50 
while (!dbHasCompleted(rs)) {
Median :4.350  Median :1.300  virginica :50 
   df <- dbFetch(rs, n = 10)
Mean  :3.758  Mean  :1.199                 
   print(nrow(df))
3rd Qu.:5.100  3rd Qu.:1.800                 
}
Max.  :6.900  Max.  :2.500                 


R> quit
dbClearResult(rs)
Connection closed by foreign host.
</pre>
</pre>


=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
'''Multiple parameterised queries''':
Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See [[R#An_example_from_Bioconductor_workshop|Example from Bioconductor workshop]].
{{Pre}}
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)
</pre>
 
'''Statements''':
{{Pre}}
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
#> [1] 0
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</pre>
 
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]
 
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===


See my [[Rserve]] page.
=== [[MySQL#Use_through_R|RMySQL]] ===
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.


=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
=== MongoDB ===
* http://www.r-bloggers.com/r-and-mongodb/
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


=== [http://rdotnet.codeplex.com/ R.NET] ===
=== odbc ===


=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
=== RODBC ===
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.


Terminal
=== DBI ===
<syntaxhighlight lang='bash'>
# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk
</syntaxhighlight>
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
<pre>
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
</pre>
* And then run '''sudo ldconfig'''


Now go back to R
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
<syntaxhighlight lang='rsplus'>
* To use databases with dplyr, you need to first install dbplyr
install.packages("rJava")
* https://db.rstudio.com/dplyr/
</syntaxhighlight>
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
Done!
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html


If above does not work, a simple way is by (under Ubuntu) running
'''Create a new SQLite database''':
<pre>
{{Pre}}
sudo apt-get install r-cran-rjava
surveys <- read.csv("data/surveys.csv")
</pre>
plots <- read.csv("data/plots.csv")
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


=== RCaller ===
my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)


=== RApache ===
copy_to(my_db, surveys)
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf
copy_to(my_db, plots)
my_db
</pre>


=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
'''Connect to a database''':
Provides hash-bang (#!) capability for R
{{Pre}}
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


FAQs:
library(dbplyr)
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
library(dplyr)
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
mammals <- src_sqlite("portal_mammals.sqlite")
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
</pre>
<syntaxhighlight lang='bash'>
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"


-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
'''Querying the database with the SQL syntax''':
                                              # Example: R -q -e "date()"
{{Pre}}
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
</pre>


-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
'''Querying the database with the dplyr syntax''':
                                              # It won't show the startup message when it is used in the command line.
{{Pre}}
                                              # Example: Rscript -e "date()"
surveys <- tbl(mammals, "surveys")
</syntaxhighlight>
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)


We can install littler using two ways.
show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
* install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ''~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r''). You need to create a soft link in order to access it globally.
</pre>
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.


After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The [http://dirk.eddelbuettel.com/code/littler.examples.html web version of examples] does not have the TOC.
'''Simple database queries''':
{{Pre}}
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</pre>


'''r''' was not meant to run interactively like '''R'''. See ''man r''.
'''Laziness''' (instruct R to stop being lazy):
{{Pre}}
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</pre>


=== RInside: Embed R in C++ ===
'''Complex database queries''':
See [[R#RInside|RInside]]
{{Pre}}
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table


(''From RInside documentation'') The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.
surveys # The plot_id column also features in the surveys table


The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
# Join databases method 1
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()
</pre>


To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.
=== NoSQL ===
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]


To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
== Github ==
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>


The real build process looks like (check <Makefile> for completeness)
=== R source  ===
<pre>
https://github.com/wch/r-sourceDaily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include  \
    rinside_sample0.cpp \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0
</pre>


Hello World example of embedding R in C++.
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
<pre>
#include <RInside.h>                    // for the embedded R via RInside


int main(int argc, char *argv[]) {
=== R packages (only) source (metacran) ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.


    RInside R(argc, argv);              // create an embedded R instance
=== Bioconductor packages source ===
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>


    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
=== Send local repository to Github in R by using reports package ===
http://www.youtube.com/watch?v=WdOI_-aZV0Y
 
=== My collection ===
* https://github.com/arraytools
* https://gist.github.com/4383351 heatmap using leukemia data
* https://gist.github.com/4382774 heatmap using sequential data
* https://gist.github.com/4484270 biocLite


    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
=== How to download ===


    exit(0);
Clone ~ Download.
}
* Command line
<pre>
git clone https://gist.github.com/4484270.git
</pre>
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


The above can be compared to the Hello world example in Qt.
* Within R
<pre>
library(devtools)
source_gist("4484270")
</pre>
or
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
<pre>
#include <QApplication.h>
library(RJSONIO)
#include <QPushButton.h>
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})
</pre>


int main( int argc, char **argv )
=== Jekyll ===
{
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
    QApplication app( argc, argv );


    QPushButton hello( "Hello world!", 0 );
== Connect R with Arduino ==
    hello.resize( 100, 30 );
* https://zhuhao.org/post/connect-arduino-chips-with-r/
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
* http://bio7.org/?p=2049
* http://www.rforge.net/Arduino/svn.html
 
== Android App ==
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]
 
== Common plots tips ==
=== Create an empty plot ===
'''plot.new()'''   


    app.setMainWidget( &hello );
=== Overlay plots ===
    hello.show();
[https://finnstats.com/index.php/2021/08/15/how-to-overlay-plots-in-r/ How to Overlay Plots in R-Quick Guide with Example].  
<pre>
#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)
</pre>


    return app.exec();
=== Save the par() and restore it ===
'''Example 1''': Don't use old.par <- par() directly. no.readonly = FALSE by default. * The '''`no.readonly = TRUE`''' argument in the [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par par()] function in R is used to get the full list of graphical parameters '''that can be restored'''.
* When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
* If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
<pre>
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)
</pre>
'''Example 2''': Use it inside a function with the [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/on.exit on.exit(0] function.
<pre>
ex <- function() {
  old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
  on.exit(par(old.par))
  ## ... do lots of par() settings and plots
  ## ...
  invisible() #-- now,  par(old.par)  will be executed
}
}
</pre>
'''Example 3''': It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.
<pre>
ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
</pre>
<pre>
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar
</pre>
</pre>


=== [http://www.rfortran.org/ RFortran] ===
=== Grouped boxplots ===
RFortran is an open source project with the following aim:
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
** 'at' parameter in boxplot() to change the equal spaced boxplots
** embed par(mar=) in boxplot()
** mtext(line=) to solve the problem the xlab overlapped with labels.
* [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others ggplot2 approach] (Hint: '''facet_grid''' is used)


''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
=== [https://www.samruston.co.uk/ Weather Time Line] ===
The plot looks similar to a boxplot though it is not. See a [https://www.samruston.co.uk/images/screens/screen_2.png screenshot] on Android by [https://www.samruston.co.uk/ Sam Ruston].


It works only on Windows platform with Microsoft Visual Studio installed:(
=== Horizontal bar plot ===
 
{{Pre}}
== Call R from other languages ==
library(ggplot2)
=== C ===
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change") 
</pre>


Error: [https://stackoverflow.com/questions/43662542/not-resolved-from-current-namespace-error-when-calling-c-routines-from-r “not resolved from current namespace” error, when calling C routines from R]
[[:File:Ggplot2bar.svg]]


Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
=== Include bar values in a barplot ===
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
* [http://stackoverflow.com/questions/12481430/how-to-display-the-frequency-at-the-top-of-each-factor-in-a-barplot-in-r barplot(), text() and axis()] functions. The data can be from a table() object.
* [https://stackoverflow.com/questions/11938293/how-to-label-a-barplot-bar-with-positive-and-negative-bars-with-ggplot2 How to label a barplot bar with positive and negative bars with ggplot2]


=== JRI ===
Use text().  
http://www.rforge.net/JRI/


=== ryp2 ===
Or use geom_text() if we are using the ggplot2 package. See an example [http://dsgeek.com/2014/09/19/Customizingggplot2charts.html here] or [https://rpubs.com/escott8908/RGC_Ch3_Gar_Graphs this].
http://rpy.sourceforge.net/rpy2.html


== Create a standalone Rmath library ==
For stacked barplot, see [http://t-redactyl.io/blog/2016/01/creating-plots-in-r-using-ggplot2-part-4-stacked-bar-plots.html this] post.
R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of [http://cran.r-project.org/doc/manuals/R-admin.html#The-standalone-Rmath-library R-admin manual].


Here is my experience based on R 3.0.2 on Windows OS.
=== Grouped barplots ===
 
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
{{Pre}}
Suppose we have downloaded R source code and build R from its source. See [[R#Build_R_from_its_source|Build_R_from_its_source]]. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.
library(ggplot2)
<pre>
# mydata <- data.frame(OUTGRP, INGRP, value)
cd C:\R\R-3.0.2\src\nmath\standalone
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
make -f Makefile.win
      geom_bar(position="dodge", stat="identity")
</pre>
* https://datascienceplus.com/building-barplots-with-error-bars/. The error bars define 2 se (95% interval) for the black-and-white version and 1 se (68% interval) for ggplots. Be careful.
{{Pre}}
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</pre>
</pre>
* [http://stackoverflow.com/questions/27466035/adding-values-to-barplot-of-table-in-r two bars in one factor] (stack). The data can be a 2-dim matrix with numerical values.
* [http://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r two bars in one factor], [https://stats.stackexchange.com/questions/14118/drawing-multiple-barplots-on-a-graph-in-r Drawing multiple barplots on a graph in R] (next to each other)
** [https://datascienceplus.com/building-barplots-with-error-bars/ Include error bars]
* [http://bl.ocks.org/patilv/raw/7360425/ Three variables] barplots
* [https://peltiertech.com/stacked-bar-chart-alternatives/ More alternatives] (not done by R)


=== Use Rmath library in our code ===
=== Math expression ===
<pre>
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
* Some cases
# It is not LD_LIBRARY_PATH in above.
** Use [https://www.rdocumentation.org/packages/base/versions/3.6.0/topics/expression expression()] function
** Don't need the backslash; use ''eta'' instead of ''\eta''. ''eta'' will be recognized as a special keyword in expression()
** Use parentheses instead of curly braces; use ''hat(eta)'' instead of ''hat{eta}''
** Summary: use expression(hat(eta)) instead of expression(\hat{\eta})
** [] means subscript, while ^ means superscript. See [https://statisticsglobe.com/add-subscript-and-superscript-to-plot-in-r Add Subscript and Superscript to Plot in R]
** Spacing can be done with ~.
** Mix math symbols and text using paste()
** Using substitute() and paste() if we need to substitute text (this part is advanced)
{{Pre}}
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")
 
# Superscript
plot(1:10, main = expression("My Title"^2))
# Subscript
plot(1:10, main = expression("My Title"[2])) 


# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
# Expressions with Spacing
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# '~' is to add space and '*' is to squish characters together
# It is OK to save the cpp file under any directory.
plot(1:10, xlab= expression(Delta * 'C'))
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
    ylab = expression(phi^{rho + a} * z * w),
    main = "Pure Expressions with Spacing")


# Force to link against the static library <libRmath.a>
# Expressions with Text
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
plot(x,y,
# OR
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe
    ylab = expression(paste("Here is some text of ", phi^{rho})),  
    main = "Expressions with Text")


# Force to link against dynamic library <Rmath.dll>
# Substituting Expressions
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
plot(x,y,
</pre>
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
<pre>
    main = "Substituted Expressions")
c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>
</pre>


Below is the cpp program <RmathEx1.cpp>.
=== Impose a line to a scatter plot ===
<pre>
* abline + lsfit # least squares
//RmathEx1.cpp
{{Pre}}
#define MATHLIB_STANDALONE
plot(cars)
#include <iostream>
abline(lsfit(cars[, 1], cars[, 2]))
#include "Rmath.h"
# OR
abline(lm(cars[,2] ~ cars[,1]))
</pre>
* abline + line # robust line fitting
{{Pre}}
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
</pre>
* lines
{{Pre}}
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')
</pre>


using std::cout; using std::cin; using std::endl;
=== How to actually make a quality scatterplot in R: axis(), mtext() ===
[https://www.r-bloggers.com/2021/08/how-to-actually-make-a-quality-scatterplot-in-r/ How to actually make a quality scatterplot in R]


int main()
=== 3D scatterplot ===
{
* [http://sthda.com/english/wiki/scatterplot3d-3d-graphics-r-software-and-data-visualization Scatterplot3d: 3D graphics - R software and data visualization]. [https://stackoverflow.com/a/24510286 how to add legend to scatterplot3d in R] and consider '''xpd=TRUE'''.
  double x1, x2;
* [[R_web#plotly|R web > plotly]]
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


  cout << "Prob(Z <= " << x1 << ") = " <<
=== Rotating x axis labels for barplot ===
    pnorm(x1, 0, 1, 1, 0)  << endl;
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
{{Pre}}
    pchisq(x2, 1, 1, 0) << endl;
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
  return 0;
}
</pre>
</pre>


== Calling R.dll directly ==
=== Set R plots x axis to show at y=0 ===
See Chapter 8.2.2 of [http://cran.r-project.org/doc/manuals/R-exts.html#Calling-R_002edll-directly|Writing R Extensions]. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
{{Pre}}
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</pre>


== Create HTML report ==
=== Different colors of axis labels in barplot ===
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor.
See [https://stackoverflow.com/questions/18839731/vary-colors-of-axis-labels-in-r-based-on-another-variable Vary colors of axis labels in R based on another variable]


=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.
{{Pre}}
tN <- table(Ni <- stats::rpois(100, lambda = 5))
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
</pre>


* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
{{Pre}}
barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
</pre>


=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
=== Use text() to draw labels on X/Y-axis including rotation ===
* https://github.com/renkun-ken/formattable
* adj = 1 means top/right alignment.  For left-bottom alignment, set adj = 0. The default is to center the text. [[https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text ?text]
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]
** par("usr") is determined *after* a plot has been created
 
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
=== [https://github.com/crubba/htmltab htmltab] package ===
* https://datascienceplus.com/building-barplots-with-error-bars/
This package is NOT used to CREATE html report but EXTRACT html table.
{{Pre}}
par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</pre>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/
 
=== Vertically stacked plots with the same x axis ===
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in


=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
=== Include labels on the top axis/margin: axis() and mtext() ===
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.
<pre>
plot(1:4, rnorm(4), axes = FALSE)
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3
</pre>
See also [[#15_Questions_All_R_Users_Have_About_Plots| 15_Questions_All_R_Users_Have_About_Plots]]


== Create academic report ==
This can be used to annotate each plot with the script name, date, ...
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.
<pre>
mtext(text=paste("Prepared on", format(Sys.time(), "%d %B %Y at %H:%M")),
      adj=.99,  # text align to right
      cex=.75, side=3, las=1, line=2)
</pre>


== Create pdf and epub files ==
ggplot2 uses '''breaks''' instead of '''at''' parameter. See [[Ggplot2#Add_axis_on_top_or_right_hand_side|ggplot2 &rarr; Add axis on top or right hand side]], [[Ggplot2#ggplot2::scale_-_axes.2Faxis.2C_legend|ggplot2 &rarr; scale_x_continus(name, breaks, labels)]] and the [https://ggplot2.tidyverse.org/reference/scale_continuous.html scale_continuous documentation].
<syntaxhighlight lang='rsplus'>
# Idea:
#        knitr        pdflatex
#  rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
</syntaxhighlight>
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
<syntaxhighlight lang='bash'>
git clone https://github.com/yihui/knitr-examples.git
</syntaxhighlight>
* <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run '''sudo apt-get install texlive-fonts-recommended''' to install missing fonts. It works!


To see a real example, check out [http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html DESeq2] package (inst/doc subdirectory). In addition to DESeq2, I also need to install '''DESeq, BiocStyle, airway, vsn, gplots''', and '''pasilla''' packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.
=== Legend tips ===
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]


Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
<syntaxhighlight lang='bash'>
{{Pre}}
# Idea:
plot(rnorm(100))
#       knitr        pandoc
# op <- par(cex=2)
#  rmd -------> md ----------> pdf
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)
</pre>


git clone https://github.com/yihui/knitr-examples.git
'''legend inset'''. Default is 0. % (from 0 to 1) to draw the legend away from x and y axis. The inset argument with [https://stackoverflow.com/a/10528078 negative values moves the legend outside the plot].
cd knitr-examples
<pre>
R -e "library(knitr); knit('001-minimal.Rmd')"
legend("bottomright", inset=.05, )
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</pre>
</syntaxhighlight>


To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
'''legend without a box'''
<syntaxhighlight lang='rsplus'>
<pre>
# Idea:
legend(, bty = "n")
#        knitr        pandoc
</pre>
#  rnw -------> tex ----------> markdown or epub


library(knitr)
'''Add a legend title'''
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</syntaxhighlight>
<pre>
<pre>
## Windows OS, epub cannot be built
legend(, title = "")
pandoc:
Error:
"source" (line 41, column 7):
unexpected "k"
expecting "{document}"
 
## Linux OS, epub missing figures and R codes.
## First install texlive base and extra packages
## sudo apt-get install texlive-latex-base texlive-latex-extra
pandoc: Could not find media `figure/SchwederSpjotvoll-1', skipping...
pandoc: Could not find media `figure/sortedP-1', skipping...
pandoc: Could not find media `figure/figHeatmap2c-1', skipping...
pandoc: Could not find media `figure/figHeatmap2b-1', skipping...
pandoc: Could not find media `figure/figHeatmap2a-1', skipping...
pandoc: Could not find media `figure/plotCountsAdv-1', skipping...
pandoc: Could not find media `figure/plotCounts-1', skipping...
pandoc: Could not find media `figure/MA-1', skipping...
pandoc: Could not find media `figure/MANoPrior-1', skipping...
</pre>
</pre>
The problems are at least
* figures need to be generated under the same directory as the source code
* figures cannot be in the format of pdf (DESeq2 generates both pdf and png files format)
* missing R codes


Convert tex to epub
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion


=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
=== Superimpose a density plot or any curves ===
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
Use '''lines()'''.


* https://rmarkdown.rstudio.com/lesson-7.html
Example 1
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
{{Pre}}
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
plot(cars, main = "Stopping Distance versus Speed")
* https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/
lines(stats::lowess(cars))
* [https://cran.r-project.org/web/packages/kableExtra/vignettes/awesome_table_in_html.html kableExtra] package
 
== Create Word report ==


=== knitr + pandoc ===
plot(density(x), col = "#6F69AC", lwd = 3)
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
lines(density(y), col = "#95DAC1", lwd = 3)
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
lines(density(z), col = "#FFEBA1", lwd = 3)
* http://rmarkdown.rstudio.com/articles_docx.html
 
It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.
<pre>
# Idea:
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</pre>
</pre>
and then
<pre>
FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
</pre>
Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.


Another way is
Example 2
<pre>
{{Pre}}
library(pander)
require(survival)
name = "demo"
n = 10000
knit(paste0(name, ".Rmd"), encoding = "utf-8")
beta1 = 2; beta2 = -1
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
lambdaT = 1 # baseline hazard
lambdaC = 2  # hazard of censoring
set.seed(1234)
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))
C <- rweibull(n, shape=1, scale=lambdaC) 
time = pmin(T,C)
status <- 1*(T <= C)
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1),  
    ylab="Survival probability",
    main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
</pre>
</pre>


Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:
Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".
* A pdf file: pandoc -s report.md -t latex -o report.pdf
 
* A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
=== log scale ===
* Openoffice: pandoc report.md -o report.odt
If we set y-axis to use log-scale, then what we display is the value log(Y) or log10(Y) though we still label the values using the input. For example, when we plot c(1, 10, 100) using the log scale, it is like we draw log10(c(1, 10, 100)) = c(0,1,2) on the plot but label the axis using the true values c(1, 10, 100).
* Word docx: pandoc report.md -o report.docx
 
[[:File:Logscale.png]]


We can also create the epub file for reading on Kobo ereader. For example, download [https://gist.github.com/jeromyanglim/2716336 this file] and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!
=== Custom scales ===
<pre>
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
knit("example.Rmd")
pandoc("example.md", format="epub")
</pre>


PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
== Time series ==
<pre>
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
> pandoc("Rmd_to_Epub.md", format="epub")
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
 
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
=== Time series stock price plot ===
Error in (function (input, format, ext, cfg) : conversion failed
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
In addition: Warning message:
* [https://datascienceplus.com/visualize-your-portfolios-performance-and-generate-a-nice-report-with-r/ Visualize your Portfolio’s Performance and Generate a Nice Report with R]
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
* https://timelyportfolio.github.io/rCharts_time_series/history.html
 
{{Pre}}
library(quantmod)
getSymbols("AAPL")
getSymbols("IBM") # similar to AAPL
getSymbols("CSCO") # much smaller than AAPL, IBM
getSymbols("DJI") # Dow Jones, huge
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
    col='orange', subset = '2017::2017-08')
 
tail(Cl(DJI))
</pre>
</pre>


=== pander ===
=== tidyquant: Getting stock data ===
Try pandoc[1] with a minimal reproducible example, you might give a try to my "[http://cran.r-project.org/web/packages/pander/ pander]" package [2] too:
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]


<pre>
=== Timeline plot ===
library(pander)
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R
            output = tempfile(), convert = 'docx')
</pre>
Where the content of the "minimal.brew" file is something you might have
got used to with Sweave - although it's using "brew" syntax instead. See
the examples of pander [3] for more details. Please note that pandoc should
be installed first, which is pretty easy on Windows.


# http://johnmacfarlane.net/pandoc/
=== Clockify ===
# http://rapporter.github.com/pander/
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]
# http://rapporter.github.com/pander/#examples


=== R2wd ===
== Circular plot ==
Use [http://cran.r-project.org/web/packages/R2wd/ R2wd] package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.  
* http://freakonometrics.hypotheses.org/20667 which uses [https://cran.r-project.org/web/packages/circlize/ circlize] package; see also the '''ComplexHeatmap''' package.
<pre>
* https://www.biostars.org/p/17728/
> library(R2wd)
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
> wdGet()
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
    installstatconnDCOM()


This will download and install the current version of statconnDCOM
== Word cloud ==
* [http://www.sthda.com/english/wiki/text-mining-and-word-cloud-fundamentals-in-r-5-simple-steps-you-should-know Text mining and word cloud fundamentals in R : 5 simple steps you should know]
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
* [https://github.com/lepennec/ggwordcloud?s=09 ggwordcloud]


You will need a working Internet connection
== Text mining ==
because installation needs to download a file.
* [https://cran.r-project.org/web/packages/tm/index.html tm] package. It was used by [https://github.com/jtleek/swfdr/blob/master/getPvalues.R R code] of [https://doi.org/10.1093/biostatistics/kxt007 An estimate of the science-wise false discovery rate and application to the top medical literature].
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :  
  argument is of length zero
</pre>


The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
== World map ==
[https://www.enchufa2.es/archives/visualising-ssh-attacks-with-r.html Visualising SSH attacks with R] ([https://cran.r-project.org/package=rworldmap rworldmap] and [https://cran.r-project.org/package=rgeolocate rgeolocate] packages)


=== Convert from pdf to word ===
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]


=== rtf ===
=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
* [https://blog.rstudio.com/2015/05/01/rstudio-v0-99-preview-graphviz-and-diagrammer/ Graphviz and DiagrammeR]
* http://rich-iannone.github.io/DiagrammeR/,
** [http://rich-iannone.github.io/DiagrammeR/io.html#r-markdown rmarkdown]
** [http://rich-iannone.github.io/DiagrammeR/graphviz_and_mermaid.html graphviz and mermaid] doc and examples
* https://donlelek.github.io/2015-03-31-dags-with-r/
* [https://mikeyharper.uk/flowcharts-in-r-using-diagrammer/ Data-driven flowcharts in R using DiagrammeR]


=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
Package xtable will produce html output. <syntaxhighlight lang='rsplus'>print(xtable(X), type="html")</syntaxhighlight>
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams


If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.
=== DAGitty (browser-based and R package) ===
* http://dagitty.net/
* https://cran.r-project.org/web/packages/dagitty/index.html


=== [http://cran.r-project.org/web/packages/ReporteRs/index.html ReporteRs] ===
=== dagR ===
Microsoft Word, Microsoft Powerpoint and HTML documents generation from R. The source code is hosted on https://github.com/davidgohel/ReporteRs
* https://cran.r-project.org/web/packages/dagR


[https://statbandit.wordpress.com/2016/10/28/a-quick-exploration-of-reporters/ A quick exploration]
=== Gmisc ===
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]


== PDF manipulation ==
=== Concept Maps ===
[https://github.com/pridiltal/staplr staplr]
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.


== R Graphs Gallery ==
=== flow ===
* [https://www.facebook.com/pages/R-Graph-Gallery/169231589826661 Romain François]
[https://cran.r-project.org/web/packages/flow/ flow], [https://predictivehacks.com/?all-tips=how-to-draw-flow-diagrams-in-r How To Draw Flow Diagrams In R]
* [http://shinyapps.stat.ubc.ca/r-graph-catalog/ R Graph Catalog] written using R + Shiny. The source code is available on [https://github.com/jennybc/r-graph-catalog Github].
* Forest plot. See the packages [https://cran.r-project.org/web/packages/rmeta/index.html rmeta] and [https://cran.r-project.org/web/packages/forestplot/ forestplot]. The forest plot can be used to plot the quantities like relative risk (with 95% CI) in survival data.


== COM client or server ==
== Venn Diagram ==
[[Venn_diagram|Venn diagram]]


=== Client ===
== hexbin plot ==
* [https://datasciencetut.com/how-to-create-a-hexbin-chart-in-r/ How to create a hexbin chart in R]
* [https://cran.r-project.org/web/packages/hextri/index.html hextri]: Hexbin Plots with Triangles. See an example on this https://www.pnas.org/content/117/48/30266#F4 paper] about the postpi method.


[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
== Bump chart/Metro map ==
https://dominikkoch.github.io/Bump-Chart/


=== Server ===
== Amazing/special plots ==
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]
See [[Amazing_plot|Amazing plot]].


== Use R under proxy ==
== Google Analytics ==
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
=== GAR package ===
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html


== RStudio ==
== Linear Programming ==
* [https://github.com/rstudio/rstudio Github]
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]


=== rstudio.cloud ===
== Linear Algebra ==
https://rstudio.cloud/
* [https://jimskinner.github.io/post/elegant-linear-algebra-in-r-with-the-matrix-package/ Elegant linear algebra in R with the Matrix package]. Matrix package is used.
* [https://datascienceplus.com/linear-algebra-for-machine-learning-and-deep-learning-in-r/ Linear Algebra for Machine Learning and Deep Learning in R]. MASS library is used.


=== Launch RStudio ===
== Amazon Alexa ==
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html


=== Create .Rproj file ===
== R and Singularity ==
If you have an existing package that doesn't have an .Rproj file, you can use devtools::use_rstudio("path/to/package") to add it.
https://rviews.rstudio.com/2017/03/29/r-and-singularity/


With an RStudio project file, you can
== Teach kids about R with Minecraft ==
* Restore .RData into workspace at startup
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html
* Save workspace to .RData on exit
* Always save history (even if no saving .RData)
* etc


=== package search ===
== Secure API keys ==
https://github.com/RhoInc/CRANsearcher
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]


=== Git ===
== Credentials and secrets ==
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
[https://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]


== Visual Studio ==
== Hide a password ==
[http://blog.revolutionanalytics.com/2017/05/r-and-python-support-now-built-in-to-visual-studio-2017.html R and Python support now built in to Visual Studio 2017]
=== keyring package ===
* https://cran.r-project.org/web/packages/keyring/index.html
* [http://theautomatic.net/2019/06/25/how-to-hide-a-password-in-r-with-the-keyring-package/ How to hide a password in R with the Keyring package]


== List files using regular expression ==
=== getPass ===
* Extension
[https://cran.r-project.org/web/packages/getPass/README.html getPass]
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
list.files(pattern = "^Something")
</pre>


Using '''Sys.glob()"' as
== Vision and image recognition ==
<pre>
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
> Sys.glob("~/Downloads/*.txt")
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>


== Hidden tool: rsync in Rtools ==
== Creating a Dataset from an Image ==
<pre>
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe


sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
== Turn pictures into coloring pages ==
total size is 1198416  speedup is 3.71
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0


c:\Rtools\bin>
== Numerical optimization ==
</pre>
[https://cran.r-project.org/web/views/NumericalMathematics.html CRAN Task View: Numerical Mathematics], [https://cran.r-project.org/web/views/Optimization.html CRAN Task View: Optimization and Mathematical Programming]
And rsync works best when we need to sync folder.
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
binary/
binary/Eula.txt
binary/cherrytree.lnk
binary/depends64.chm
binary/depends64.dll
binary/depends64.exe
binary/mtputty.exe
binary/procexp.chm
binary/procexp.exe
binary/pscp.exe
binary/putty.exe
binary/sqlite3.exe
binary/wget.exe


sent 4115294 bytes  received 244 bytes  1175868.00 bytes/sec
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/uniroot.html uniroot]: One Dimensional Root (Zero) Finding. This is used in [http://onlinelibrary.wiley.com/doi/10.1002/sim.7178/full simulating survival data for predefined censoring rate]
total size is 8036311  speedup is 1.95
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.


c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
== Ryacas: R Interface to the 'Yacas' Computer Algebra System ==
cygwin warning:
[https://blog.ephorie.de/doing-maths-symbolically-r-as-a-computer-algebra-system-cas Doing Maths Symbolically: R as a Computer Algebra System (CAS)]
  MS-DOS style path detected: c:\users\limingc\Documents\binary\procexp.exe
  Preferred POSIX equivalent is: /cygdrive/c/users/limingc/Documents/binary/procexp.exe
  CYGWIN environment variable option "nodosfilewarning" turns off this warning.
  Consult the user's guide for more details about POSIX paths:
    http://cygwin.com/cygwin-ug-net/using.html#using-pathnames


c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
== Game ==
sending incremental file list
* [https://kbroman.org/miner_book/?s=09 R Programming with Minecraft]
binary/
* [https://cran.r-project.org/web/packages/pixelpuzzle/index.html pixelpuzzle]
binary/procexp.exe
* [https://www.rostrum.blog/2022/09/24/pixeltrix/ Interactive pixel art in R with {pixeltrix}]
* [https://rtaoist.blogspot.com/2021/03/r-shiny-maths-games-for-6-years-old.html Shiny math game]
* [https://cran.microsoft.com/web/packages/mazing/index.html mazing]: Utilities for Making and Plotting Mazes
* [https://github.com/jeroenjanssens/raylibr/blob/main/demo/snake.R snake] which is based on [https://github.com/jeroenjanssens/raylibr raylibr]


sent 1767277 bytes  received 35 bytes  3534624.00 bytes/sec
== Music ==
total size is 8036311  speedup is 4.55
* [https://flujoo.github.io/gm/ gm]. Require to install [https://musescore.org/en MuseScore], an open source and free notation software.


c:\Rtools\bin>
== SAS ==
</pre>
[https://github.com/MangoTheCat/sasMap sasMap] Static code analysis for SAS scripts


Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also http://superuser.com/questions/69620/rsync-file-permissions-on-windows
= R packages =
[[R_packages|R packages]]


== Install rgdal package (geospatial Data) on ubuntu ==
= Tricks =
Terminal
<syntaxhighlight lang='bash'>
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
</syntaxhighlight>


R
== Getting help ==
<syntaxhighlight lang='rsplus'>
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
install.packages("rgdal")
* https://stat.ethz.ch/pipermail/r-help/
</syntaxhighlight>
* https://stat.ethz.ch/pipermail/r-devel/


== Install sf package ==
== Better Coder/coding, best practices ==
I got the following error even I have installed some libraries.  
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
<pre>
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
checking GDAL version >= 2.0.1... no
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
configure: error: sf is not compatible with GDAL versions below 2.0.1
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]
</pre>
* [https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.9169?campaign=woletoc Best practices in statistical computing] Sanchez 2021
Then I follow the instruction here
<syntaxhighlight lang='bash'>
sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable


sudo apt update
== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
sudo apt install libgdal-dev
</syntaxhighlight>


== Set up Emacs on Windows ==
== Getting user's home directory ==
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
See [https://cran.r-project.org/bin/windows/base/rw-FAQ.html#What-are-HOME-and-working-directories_003f What are HOME and working directories?]
<pre>
{{Pre}}
(setq-default inferior-R-program-name
# Windows
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
normalizePath("~")  # "C:\\Users\\brb\\Documents"
</pre>
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")  # "C:/Users/brb/Documents"


== Database ==
# Mac
* https://cran.r-project.org/web/views/Databases.html
normalizePath("~")  # [1] "/Users/brb"
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # "/Users/brb"


=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
# Linux
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
normalizePath("~")  # [1] "/home/brb"
* https://github.com/rstats-db/RSQLite
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # [1] "/home/brb"
</pre>


'''Creating a new database''':
== tempdir() ==
<syntaxhighlight lang='rsplus'>
The path is a per-session temporary directory. On parallel use, R processes forked by functions such as '''mclapply''' and '''makeForkCluster''' in package '''parallel''' share a per-session temporary directory.
library(DBI)


mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
== Distinguish Windows and Linux/Mac, R.Version() ==
dbDisconnect(mydb)
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.
unlink("my-db.sqlite")


# temporary database
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
mydb <- dbConnect(RSQLite::SQLite(), "")
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
dbDisconnect(mydb)
<pre>
</syntaxhighlight>
get_os <- function(){
  sysinf <- Sys.info()
  if (!is.null(sysinf)){
    os <- sysinf['sysname']
    if (os == 'Darwin')
      os <- "osx"
  } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
</pre>
<pre>
names(R.Version())
#  [1] "platform"      "arch"          "os"            "system"       
#  [5] "status"        "major"          "minor"          "year"         
#  [9] "month"          "day"            "svn rev"        "language"     
# [13] "version.string" "nickname"
getRversion()
# [1] ‘4.3.0’
</pre>


'''Loading data''':
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
<syntaxhighlight lang='rsplus'>
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
mydb <- dbConnect(RSQLite::SQLite(), "")
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
dbWriteTable(mydb, "mtcars", mtcars)
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
dbWriteTable(mydb, "iris", iris)
* [http://blog.revolutionanalytics.com/2015/11/how-to-store-and-use-authentication-details-with-r.html How to store and use webservice keys and authentication details]
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]
* [https://rviews.rstudio.com/2017/04/19/r-for-enterprise-understanding-r-s-startup/ *R for Enterprise: Understanding R’s Startup]
* [https://support.rstudio.com/hc/en-us/articles/360047157094-Managing-R-with-Rprofile-Renviron-Rprofile-site-Renviron-site-rsession-conf-and-repos-conf *Managing R with .Rprofile, .Renviron, Rprofile.site, Renviron.site, rsession.conf, and repos.conf]


dbListTables(mydb)
If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
<pre>
R_LIBS_SITE=F:/R/library
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''. In R, run '''Sys.getenv("R_LIBS_SITE")''' or '''Sys.getenv("R_LIBS_USER")''' to query the environment variable. See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/EnvVar.html Environment Variables].
 
=== What is the best place to save Rconsole on Windows platform ===
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.


dbListFields(con, "mtcars")
My preferred settings:
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* Size: 12
* background: black
* normaltext: white
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow


dbReadTable(con, "mtcars")
and others (default options)
</syntaxhighlight>
* pagebg: white
* pagetext: navy
* highlight: DarkRed
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


'''Queries''':
A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].
<syntaxhighlight lang='rsplus'>
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
=== How R starts up ===
https://rstats.wtf/r-startup.html


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
=== startup - Friendly R Startup Configuration ===
https://github.com/henrikbengtsson/startup


res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
== Saving and loading history automatically: .Rprofile & local() ==
dbFetch(res)
<ul>
</syntaxhighlight>
<li>[http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html savehistory("filename")]. It will save everything from the beginning to the command savehistory() to a text file.
<li>'''.Rprofile''' will automatically be loaded when R has started from that directory
<li>Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
<li>'''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
<li>[http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
* You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
* At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
options(continue="  ") # default is "+ "
options(prompt="R> ", continue=" ")
options(editor="nano") # default is "vi" on Linux
# options(htmlhelp=TRUE)


'''Batched queries''':
local({r <- getOption("repos")
<syntaxhighlight lang='rsplus'>
      r["CRAN"] <- "https://cran.rstudio.com"
dbClearResult(rs)
      options(repos=r)})
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


dbClearResult(rs)
.First <- function(){
</syntaxhighlight>
# library(tidyverse)
cat("\nWelcome at", date(), "\n")
}


'''Multiple parameterised queries''':
.Last <- function(){
<syntaxhighlight lang='rsplus'>
cat("\nGoodbye at ", date(), "\n")
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
</pre>
nrow(dbFetch(rs))
<li>https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
#> [1] 4
<li>The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
dbClearResult(rs)
<li>[https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
</syntaxhighlight>
</ul>
'''Linux''' or '''Mac'''


'''Statements''':
In '''~/.profile''' or '''~/.bashrc''' I put:
<syntaxhighlight lang='rsplus'>
<pre>
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
export R_HISTFILE=~/.Rhistory
#> [1] 0
</pre>
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
In '''~/.Rprofile''' I put:
dbBind(rs, param = list(x = 4.5))
<pre>
dbGetRowsAffected(rs)
if (interactive()) {
#> [1] 4
  if (.Platform$OS.type == "unix") .First <- function() try(utils::loadhistory("~/.Rhistory"))
dbClearResult(rs)
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
</syntaxhighlight>
}
</pre>


=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
'''Windows'''
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]


=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
If you launch R by clicking its icon from Windows Desktop, the R starts in '''C:\User\$USER\Documents''' directory. So we can create a new file '''.Rprofile''' in this directory.
<pre>
if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>


=== [[MySQL#Use_through_R|RMySQL]] ===
== Disable "Save workspace image?" prompt when exit R? ==
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in R?]
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.


=== MongoDB ===
== R release versions ==
* http://www.r-bloggers.com/r-and-mongodb/
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


=== odbc ===
== getRversion() ==
<pre>
getRversion()
[1] ‘4.3.0’
</pre>


=== RODBC ===
== Detect number of running R instances in Windows ==
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
<pre>
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
INFO: No tasks are running which match the specified criteria.


=== DBI ===
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"


=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
Image Name                    PID Session Name        Session#    Mem Usage
* To use databases with dplyr, you need to first install dbplyr
============================================================================
* https://db.rstudio.com/dplyr/
Rgui.exe                      1096 Console                    1    44,712 K
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html


'''Create a new SQLite database''':
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
<syntaxhighlight lang='rsplus'>
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")


my_db_file <- "portal-database.sqlite"
Image Name                    PID Session Name        Session#    Mem Usage
my_db <- src_sqlite(my_db_file, create = TRUE)
============================================================================
 
Rserve.exe                    6108 Console                    1    381,796 K
copy_to(my_db, surveys)
</pre>
copy_to(my_db, plots)
In R, we can use
my_db
<pre>
</syntaxhighlight>
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
[1] ""                                                                           
[2] "Image Name                    PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1    44,804 K"


'''Connect to a database''':
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
<syntaxhighlight lang='rsplus'>
</pre>
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


library(dbplyr)
== Editor ==
library(dplyr)
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
mammals <- src_sqlite("portal_mammals.sqlite")
</syntaxhighlight>


'''Querying the database with the SQL syntax''':
<ul>
<syntaxhighlight lang='rsplus'>
<li>Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
</syntaxhighlight>
<pre>
(setq-default inferior-R-program-name
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>
* [https://blog.rwhitedwarf.com/post/use_emacs_for_r/ Using Emacs for R] 2022
</ul>
* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager. The new version (0.98) also provides a new feature for debugging step-by-step. See also [https://www.rstudio.com/rviews/2016/11/11/easy-tricks-you-mightve-missed/ RStudio Tricks]
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code. RStudio can also do this (see the bottom of the code panel).
* [http://rgedit.sourceforge.net/ Rgedit] which includes a feature of splitting screen into two panes and run R in the bottom panel. See [http://www.stattler.com/article/using-gedit-or-rgedit-r here].
* Komodo IDE with browser preview http://www.youtube.com/watch?v=wv89OOw9roI at 4:06 and http://docs.activestate.com/komodo/4.4/editor.html


'''Querying the database with the dplyr syntax''':
== GUI for Data Analysis ==
<syntaxhighlight lang='rsplus'>
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023
surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)


show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
=== BlueSky Statistics ===
</syntaxhighlight>
* https://www.blueskystatistics.com/Default.asp
* [https://r4stats.com/articles/software-reviews/bluesky/ A Comparative Review of the BlueSky Statistics GUI for R]


'''Simple database queries''':
=== Rcmdr ===
<syntaxhighlight lang='rsplus'>
http://cran.r-project.org/web/packages/Rcmdr/index.html. After loading a dataset, click Statistics -> Fit models. Then select Linear regression, Linear model, GLM, Multinomial logit model, Ordinal regression model, Linear mixed model, and Generalized linear mixed model. However, Rcmdr does not include, e.g. random forest, SVM, glmnet, et al.
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</syntaxhighlight>


'''Laziness''' (instruct R to stop being lazy):
=== Deducer ===
<syntaxhighlight lang='rsplus'>
http://cran.r-project.org/web/packages/Deducer/index.html
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</syntaxhighlight>


'''Complex database queries''':
=== jamovi ===
<syntaxhighlight lang='rsplus'>
* https://www.jamovi.org/
plots <- tbl(mammals, "plots")
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
plots # # The plot_id column features in the plots table


surveys # The plot_id column also features in the surveys table
== Scope ==
See
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.


# Join databases method 1
=== source() ===
plots %>%
* [https://twitter.com/henrikbengtsson/status/1563849697084809219?s=20&t=nStcqVabAQ_HvJ2FaBloNQ source() assigns to the global environment, not the calling environment, which might not be what you want/expect]. Instead, use source("file.R", local = TRUE) to avoid assigning functions and variables to the global environment.
  filter(plot_id == 1) %>%
* [[#How_to_exit_a_sourced_R_script|source()]] does not work like C's preprocessor where statements in header files will be literally inserted into the code. It does not work when you define a variable in a function but want to use it outside the function (even through '''source()''')
  inner_join(surveys) %>%
  collect()
</syntaxhighlight>


=== NoSQL ===
{{Pre}}
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
## foo.R ##
cat(ArrayTools, "\n")
## End of foo.R


== Github ==
# 1. Error
 
predict <- function() {
=== R source  ===
  ArrayTools <- "C:/Program Files" # or through load() function
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
  source("foo.R")                  # or through a function call; foo()
}
predict()  # Object ArrayTools not found


If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
# 2. OK. Make the variable global
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict() 
ArrayTools


=== R packages (only) source (metacran) ===
# 3. OK. Create a global variable
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()
</pre>


=== Bioconductor packages source ===
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>


=== Send local repository to Github in R by using reports package ===
Example 1.  
http://www.youtube.com/watch?v=WdOI_-aZV0Y
 
=== My collection ===
* https://github.com/arraytools
* https://gist.github.com/4383351 heatmap using leukemia data
* https://gist.github.com/4382774 heatmap using sequential data
* https://gist.github.com/4484270 biocLite
 
=== How to download ===
 
Clone ~ Download.
* Command line
<pre>
<pre>
git clone https://gist.github.com/4484270.git
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
</pre>
> ttt
This will create a subdirectory called '4484270' with all cloned files there.
  type JpnTest
1    a    999
2    b    999
3    c    999
4    d    999
5    e    999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a      N5
2    b    999
3    c    999
4    d    999
5    e    999
> ttt
  type JpnTest
1    a    999
2    b    999
3    c    999
4    d    999
5    e    999
</pre>
 
Example 2. [http://stackoverflow.com/questions/1236620/global-variables-in-r How can we set global variables inside a function?] The answer is to use the "<<-" operator or '''assign(, , envir = .GlobalEnv)''' function.
 
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
 
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
 
=== New environment ===
* http://adv-r.had.co.nz/Environments.html.
* [https://www.r-bloggers.com/2011/06/environments-in-r/ Environments in R]
* load(), attach(), with().  
* [https://stackoverflow.com/questions/33109379/how-to-switch-to-a-new-environment-and-stick-into-it How to switch to a new environment and stick into it?] seems not possible!
 
Run the same function on a bunch of R objects
{{Pre}}
mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(<nowiki>mye[[n]]</nowiki>)
</pre>


* Within R
Just look at the contents of rda file without saving to anywhere (?load)
<pre>
<pre>
library(devtools)
local({
source_gist("4484270")
  load("myfile.rda")
  ls()
})
</pre>
</pre>
or
Or use '''attach()''' which is a wrapper of load(). It creates an environment and slots it into the list right after the global environment, then populates it with the objects we're attaching.
First download the json file from  
{{Pre}}
https://api.github.com/users/MYUSERLOGIN/gists
attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
and then
ls(pos = 2)
##  also typically need to cleanup the search path:
detach("file:all.rda")
</pre>
If we want to read data from internet, '''load()''' works but not attach().
<pre>
con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840
</pre>
[https://stackoverflow.com/a/39621091 source() case].
<pre>
<pre>
library(RJSONIO)
myEnv <- new.env()  
x <- fromJSON("~/Downloads/gists.json")
source("some_other_script.R", local=myEnv)
setwd("~/Downloads/")
attach(myEnv, name="sourced_scripts")
gist.id <- lapply(x, "[[", "id")
search()
lapply(gist.id, function(x){
ls(2)
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
ls(myEnv)
  system(cmd)
with(myEnv, print(x))
})
</pre>
</pre>


=== Jekyll ===
=== str( , max) function ===
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
Use '''max.level''' parameter to avoid a long display of the structure of a complex R object. Use '''give.head = FALSE''' to hide the attributes. See [https://www.rdocumentation.org/packages/utils/versions/3.6.1/topics/str ?str]
 
If we use str() on a function like str(lm), it is equivalent to args(lm)


== Connect R with Arduino ==
For a complicated list object, it is useful to use the '''max.level''' argument; e.g. str(, max.level = 1)
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
* http://bio7.org/?p=2049
* http://www.rforge.net/Arduino/svn.html


== Android App ==
For a large data frame, we can use the '''tibble()''' function; e.g. mydf %>% tibble()
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]


== Common plots tips ==
=== tidy() function ===
=== Grouped boxplots ===
broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See [[Tidyverse#broom|here]].
* [http://sphaerula.com/legacy/R/boxplotTwoWay.html Box Plots of Two-Way Layout]
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
** 'at' parameter in boxplot() to change the equal spaced boxplots
** embed par(mar=) in boxplot()
** mtext(line=) to solve the problem the xlab overlapped with labels.
* [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others ggplot2 approach] (Hint: '''facet_grid''' is used)


=== [https://www.samruston.co.uk/ Weather Time Line] ===
=== View all objects present in a package, ls() ===
The plot looks similar to a boxplot though it is not. See a [https://www.samruston.co.uk/images/screens/screen_2.png screenshot] on Android by [https://www.samruston.co.uk/ Sam Ruston].
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
{{Pre}}
> devtools::load_all("mypackage")
> search()
[1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
[4] "package:stats"    "package:graphics"  "package:grDevices"
[7] "package:utils"    "package:datasets"  "package:methods"
[10] "Autoloads"        "package:base"


=== Horizontal bar plot ===
> ls("package:mypackage")
<syntaxhighlight lang='rsplus'>
[1] "_mypackage_rcpp_hello_world" "evalCpp"                     "library.dynam.unload"      
library(ggplot2)
[4] "rcpp_hello_world"           "system.file"
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
</pre>
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change"
</syntaxhighlight>


[[File:Ggplot2bar.svg|300px]]
Note that the first argument of ls() (or detach()) is used to specify the environment. It can be
* an integer (the position in the ‘search’ list);
* the character string name of an element in the search list;
* an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).


=== Include bar values in a barplot ===
== Speedup R code ==
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
* [http://stackoverflow.com/questions/12481430/how-to-display-the-frequency-at-the-top-of-each-factor-in-a-barplot-in-r barplot(), text() and axis()] functions. The data can be from a table() object.
 
* [https://stackoverflow.com/questions/11938293/how-to-label-a-barplot-bar-with-positive-and-negative-bars-with-ggplot2 How to label a barplot bar with positive and negative bars with ggplot2]
=== Profiler ===
* [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler] (Video)
* [https://github.com/atheriel/xrprof-package xrprof] package, [https://www.infoworld.com/article/3604688/top-r-tips-and-news-from-rstudio-global-2021.amp.html Top R tips and news from RStudio Global 2021]
 
== && vs & ==
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.


Use text().  
* The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
* The longer form evaluates left to right examining only the first element of each vector. The return is one value.
* '''The longer form''' evaluates left to right examining only the first element of each vector. '''Evaluation proceeds only until the result is determined.'''
* The idea of the longer form && in R seems to be the same as the && operator in linux shell; see [https://youtu.be/AVXYq8aL47Q?t=1475 here].
* [https://medium.com/biosyntax/single-or-double-and-operator-and-or-operator-in-r-442f00332d5b Single or double?: AND operator and OR operator in R]. The confusion might come from the inconsistency when choosing these operators in different languages. For example, in C, & performs bitwise AND, while && does Boolean logical AND.
* [https://www.tjmahr.com/think-of-stricter-logical-operators/ Think of && as a stricter &]


Or use geom_text() if we are using the ggplot2 package. See an example [http://dsgeek.com/2014/09/19/Customizingggplot2charts.html here] or [https://rpubs.com/escott8908/RGC_Ch3_Gar_Graphs this].
<pre>
c(T,F,T) & c(T,T,T)
# [1]  TRUE FALSE  TRUE
c(T,F,T) && c(T,T,T)
# [1] TRUE
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
</pre>
<pre>
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated
</pre>
It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.
<pre>
> foo <- function(arg, L) {
  # Suppose 'L' is meaningful only if 'arg' is provided
  #
  # Evaluate 'L' only if 'arg' is provided
  #
  if (!missing(arg) && L) {
    print("L is true")
  } else {
    print("Either arg is missing or L is FALSE")
  }
}
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"
</pre>
Other examples: '''&&''' is more flexible than '''&'''.
<pre>
nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))


For stacked barplot, see [http://t-redactyl.io/blog/2016/01/creating-plots-in-r-using-ggplot2-part-4-stacked-bar-plots.html this] post.
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
</pre>


=== Grouped barplots ===
== for-loop, control flow ==
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)<syntaxhighlight lang='rsplus'>
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Control ?Control]
library(ggplot2)
* '''next''' can be used to skip the rest of the inner-most loop
# mydata <- data.frame(OUTGRP, INGRP, value)
* [https://www.programiz.com/r/ifelse-function ifelse() Function]
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
      geom_bar(position="dodge", stat="identity")
</syntaxhighlight>
* https://datascienceplus.com/building-barplots-with-error-bars/. The error bars define 2 se (95% interval) for the black-and-white version and 1 se (68% interval) for ggplots. Be careful.<syntaxhighlight lang='rsplus'>
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</syntaxhighlight>
* [http://stackoverflow.com/questions/27466035/adding-values-to-barplot-of-table-in-r two bars in one factor] (stack). The data can be a 2-dim matrix with numerical values.
* [http://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r two bars in one factor], [https://stats.stackexchange.com/questions/14118/drawing-multiple-barplots-on-a-graph-in-r Drawing multiple barplots on a graph in R] (next to each other)
** [https://datascienceplus.com/building-barplots-with-error-bars/ Include error bars]
* [http://bl.ocks.org/patilv/raw/7360425/ Three variables] barplots
* [https://peltiertech.com/stacked-bar-chart-alternatives/ More alternatives] (not done by R)


=== Math expression ===
== Vectorization ==
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* https://andyphilips.github.io/blog/2017/08/16/mathematical-symbols-in-r-plots.html  
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
** Use [https://www.rdocumentation.org/packages/base/versions/3.6.0/topics/expression expression()] function
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
** Don't need the backslash; use ''eta'' instead of ''\eta''. ''eta'' will be recognized as a special keyword in expression()
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html  
** Use parentheses instead of curly braces; use ''hat(eta)'' instead of ''hat{eta}''
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
** Summary: use expression(hat(eta)) instead of expression(\hat{\eta})
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
** [] means subscript, while ^ means superscript
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
** Spacing can be done with ~.
* [https://github.com/henrikbengtsson/matrixstats matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors). E.g. col / rowMedians(), col / rowRanks(), and col / rowSds(). [https://github.com/HenrikBengtsson/matrixStats/wiki/Benchmark-reports Benchmark reports].
** Mix math symbols and text using paste()
** Using substitute() and paste() if we need to substitute text (this part is advanced)
: <syntaxhighlight lang='rsplus'>
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")


# Expressions with Spacing
=== sapply vs vectorization ===
# '~' is to add space and '*' is to squish characters together
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
plot(1:10, xlab= expression(Delta * 'C'))
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
    ylab = expression(phi^{rho + a} * z * w),
    main = "Pure Expressions with Spacing")


# Expressions with Text
=== lapply vs for loop ===
plot(x,y,
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
* https://code-examples.net/en/q/286e03a
    ylab = expression(paste("Here is some text of ", phi^{rho})),
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]
    main = "Expressions with Text")


# Substituting Expressions
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
plot(x,y,
split() can be used to split a vector, columns or rows. See [https://stackoverflow.com/questions/3302356/how-to-split-a-data-frame How to split a data frame?]
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),  
* Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
{{Pre}}
    main = "Substituted Expressions")
split(x, f, drop = FALSE, …)
</syntaxhighlight>
</pre>
* [https://stackoverflow.com/a/3321659 Split a vector into chunks]. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the '''split() + lapply() + do.call()''' or '''split() + sapply()''' operations.
<pre>
d <- 1:10
chunksize <- 4
ceiling(1:10/4)
# [1] 1 1 1 1 2 2 2 2 3 3
split(d, ceiling(seq_along(d)/chunksize))
# $`1`
# [1] 1 2 3 4
#
# $`2`
# [1] 5 6 7 8
#
# $`3`
# [1]  9 10
do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) )  
#  1  2  3
# 10 26 19


=== Impose a line to a scatter plot ===
# bigmemory vignette
* abline + lsfit # least squares
planeindices <- split(1:nrow(x), x[,'TailNum'])
: <syntaxhighlight lang='rsplus'>
planeStart <- sapply(planeindices,
plot(cars)
                    function(i) birthmonth(x[i, c('Year','Month'),
abline(lsfit(cars[, 1], cars[, 2]))
                                            drop=FALSE]))
# OR
</pre>
abline(lm(cars[,2] ~ cars[,1]))
* Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.  
</syntaxhighlight>
{{Pre}}
* abline + line # robust line fitting
split(mtcars,mtcars$cyl)
: <syntaxhighlight lang='rsplus'>
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
</syntaxhighlight>
* lines
: <syntaxhighlight lang='rsplus'>
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')
</syntaxhighlight>
=== Rotating x axis labels for barplot ===
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
<syntaxhighlight lang='rsplus'>
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</syntaxhighlight>


=== Set R plots x axis to show at y=0 ===
split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))  # split.data.frame() works for matrices
<syntaxhighlight lang='rsplus'>
</pre>
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
* Split columns of a data frame/matrix.
</syntaxhighlight>
{{Pre}}
 
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
=== Different colors of axis labels in barplot ===
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
See [https://stackoverflow.com/questions/18839731/vary-colors-of-axis-labels-in-r-based-on-another-variable Vary colors of axis labels in R based on another variable]
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
</pre>
* split() + sapply() to merge columns. See below [[#Mean_of_duplicated_columns:_rowMeans.3B_compute_Means_by_each_row|Mean of duplicated columns]] for more detail.
* split() + sapply() to split a vector. See [https://www.rdocumentation.org/packages/genefilter/versions/1.54.2/topics/nsFilter nsFilter()] function which can remove duplicated probesets/rows using unique Entrez Gene IDs ('''genefilter''' package). The source code of [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/nsFilter.R#L224 nsFilter()] and [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/all.R#L170 findLargest()].
{{Pre}}
tSsp = split.default(testStat, lls)
# testStat is a vector of numerics including probeset IDs as names
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.


Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
sapply(tSsp, function(x) names(which.max(x)))  
<syntaxhighlight lang='rsplus'>
# return a vector of probset IDs of length of unique entrez IDs
tN <- table(Ni <- stats::rpois(100, lambda = 5))
</pre>
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
</syntaxhighlight>


Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
=== strsplit and sapply ===
<syntaxhighlight lang='rsplus'>
{{Pre}}
barplot(tN, col = rainbow(20), axisnames = F)
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
> strsplit(namedf, " ")
</syntaxhighlight>
[[1]]
[1] "John" "ABC"  


=== Use [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text text()] to draw labels on X/Y-axis including rotation ===
[[2]]
* adj = 1 means top/rigth alignment. The default is to center the text.
[1] "Mary" "CDE"  
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
** par("usr") is determined *after* a plot has been created
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
* https://datascienceplus.com/building-barplots-with-error-bars/
<syntaxhighlight lang='rsplus'>
par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</syntaxhighlight>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/


=== Vertically stacked plots with the same x axis ===
[[3]]
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in
[1] "Kat" "FGH"


=== Increase/decrease legend font size ===
> sapply(strsplit(namedf, " "), "[", 1)
https://stackoverflow.com/a/36842578
[1] "John" "Mary" "Kat"
<syntaxhighlight lang='rsplus'>
> sapply(strsplit(namedf, " "), "[", 2)
plot(rnorm(100))
[1] "ABC" "CDE" "FGH"
op <- par(cex=2)
</pre>
legend("topleft", legend = 1:4, col=1:4, pch=1)
par(op)
</syntaxhighlight>


=== Superimpose a density plot or any curves ===
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
Use '''lines()'''.  
<ul>
<li>[https://stackoverflow.com/questions/35925529/reduce-columns-of-a-matrix-by-a-function-in-r Reduce columns of a matrix by a function in R]. To use rowMedians() instead of rowMeans(), we need to install [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats] from CRAN.
<syntaxhighlight lang='r'>
set.seed(1)
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res  # notice the sorting of columns
      a  b  c
[1,] 46  1 31
[2,] 47 12 32
[3,] 48 13 33
[4,] 49 14 34
[5,] 50 15 35
[6,] 51 16 36
[7,] 52 17 37
[8,] 53 18 38
[9,] 54 19 39
[10,] 55 20 40


Example 1
# vapply() is safter than sapply().
<syntaxhighlight lang='rsplus'>
# The 3rd arg in vapply() is a template of the return value.
plot(cars, main = "Stopping Distance versus Speed")
res2 <- vapply(split(1:ncol(x), colnames(x)),
lines(stats::lowess(cars))
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
              rep(0, nrow(x)))
</syntaxhighlight>
</syntaxhighlight>
</li>
<li>[https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/colSums colSums, rowSums, colMeans, rowMeans] (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster.
{{Pre}}
rowMeans(x, na.rm=T)
# [1] 31 27 28 29 30 31 32 33 34 35


Example 2
apply(x, 1, mean, na.rm=T)
<syntaxhighlight lang='rsplus'>
# [1] 31 27 28 29 30 31 32 33 34 35
require(survival)
</pre>
n = 10000
</li>
beta1 = 2; beta2 = -1
<li>[https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
lambdaT = 1 # baseline hazard
</li>
lambdaC = 2  # hazard of censoring
<li>[https://www.statforbiology.com/2020/stat_r_tidyverse_columnwise/ From ''for()'' loops to the ''split-apply-combine'' paradigm for column-wise tasks: the transition for a dinosaur]
set.seed(1234)
</li>
x1 = rnorm(n,0)
</ul>
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))
C <- rweibull(n, shape=1, scale=lambdaC) 
time = pmin(T,C) 
status <- 1*(T <= C)  
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1),
    ylab="Survival probability",
    main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
</syntaxhighlight>


=== Custom scales ===
=== Mean of duplicated rows: colMeans and rowsum ===
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/colSums colMeans(x, na.rm = FALSE, dims = 1)], take mean per columns & sum over rows. It returns a vector. Other similar idea functions include '''colSums, rowSums, rowMeans'''.
{{Pre}}
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
rownames(x) <- c(rep("b", 2), rep("c", 3), rep("d", 4), "a") # move 'a' to the last
res <- sapply(split(1:nrow(x), rownames(x)),
              function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
res  # still a matrix, rows are ordered
#  [,1] [,2] [,3] [,4] [,5] [,6]
# a 10.0 20.0 30.0 40.0 50.0 60.0
# b  1.5 12.0 22.0 31.5 41.5 51.5
# c  4.0 14.0 24.0 34.0 44.0 54.0
# d  7.5 17.5 27.5 37.5 47.5 57.5
table(rownames(x))
# a b c d
# 1 2 3 4


== Time series ==
aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
#  Group.1  V1  V2  V3  V4  V5  V6
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
# 1      a 10.0 20.0 30.0 40.0 50.0 60.0
# 2      b  1.5 12.0 22.0 31.5 41.5 51.5
# 3      c  4.0 14.0 24.0 34.0 44.0 54.0
# 4      d  7.5 17.5 27.5 37.5 47.5 57.5
</pre>
<li>[[Arraytools#Reducing_multiple_probes.2Fprobe_sets_to_one_per_gene_symbol|Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays]]


=== Time series stock price plot ===
</li>
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/rowsum rowsum(x, group, reorder = TRUE, …)]. Sum over rows. It returns a matrix. This is very special. It's not the same as rowSums. There is no "colsum" function. ''It has the speed advantage over sapply+colSums OR aggregate.''
* [https://datascienceplus.com/visualize-your-portfolios-performance-and-generate-a-nice-report-with-r/ Visualize your Portfolio’s Performance and Generate a Nice Report with R]
{{Pre}}
* https://timelyportfolio.github.io/rCharts_time_series/history.html
group <- rownames(x)
 
rowsum(x, group, na.rm=T)/as.vector(table(group))
<syntaxhighlight lang='rsplus'>
#  [,1] [,2] [,3] [,4] [,5] [,6]
library(quantmod)
# a 10.0 20.0 30.0 40.0 50.0 60.0
getSymbols("AAPL")
# b  1.5  6.0 11.0 31.5 41.5 51.5
getSymbols("IBM") # similar to AAPL
# c  4.0 14.0 24.0 34.0 44.0 54.0
getSymbols("CSCO") # much smaller than AAPL, IBM
# d  7.5 17.5 27.5 37.5 47.5 57.5
getSymbols("DJI") # Dow Jones, huge
</pre>
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
</li>
     col='orange', subset = '2017::2017-08')
</ul>
 
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
tail(Cl(DJI))
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
</syntaxhighlight>
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
 
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
=== Timeline plot ===
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
<ul>
<li>'''by()''' function. [https://thomasadventure.blog/posts/calculating-change-from-baseline-in-r/ Calculating change from baseline in R]
</li>
<li>See [https://finnstats.com/index.php/2021/06/20/aggregate-function-in-r/ '''aggregate''' Function in R- A powerful tool for data frames] & [https://finnstats.com/index.php/2021/06/01/summarize-in-r-data-summarization-in-r/ summarize in r, Data Summarization In R] </li>
<li>[http://www.statmethods.net/management/aggregate.html aggregate()] function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. [http://www.win-vector.com/blog/2015/10/dont-use-statsaggregate/ Don't use aggregate] post.
{{Pre}}
> attach(mtcars)
dim(mtcars)
[1] 32 11
> head(mtcars)
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
Mazda RX4        21.0  6  160 110 3.90 2.620 16.46  0  1    4    4
Mazda RX4 Wag    21.0  6  160 110 3.90 2.875 17.02  0  1    4    4
Datsun 710        22.8  4  108  93 3.85 2.320 18.61  1  1    4    1
Hornet 4 Drive    21.4  6  258 110 3.08 3.215 19.44  1  0    3    1
Hornet Sportabout 18.7  8  360 175 3.15 3.440 17.02  0  0    3    2
Valiant          18.1  6  225 105 2.76 3.460 20.22  1  0    3    1
> with(mtcars, table(cyl, vs))
  vs
cyl  0  1
  4  1 10
  6  3  4
  8 14  0
> aggdata <-aggregate(mtcars, by=list(cyl,vs), FUN=mean, na.rm=TRUE)
> print(aggdata)
  Group.1 Group.2      mpg cyl  disp      hp    drat      wt    qsec vs
1      4      0 26.00000  4 120.30  91.0000 4.430000 2.140000 16.70000  0
2      6      0 20.56667  6 155.00 131.6667 3.806667 2.755000 16.32667  0
3      8      0 15.10000  8 353.10 209.2143 3.229286 3.999214 16.77214  0
4      4      1 26.73000  4 103.62  81.8000 4.035000 2.300300 19.38100  1
5      6      1 19.12500  6 204.55 115.2500 3.420000 3.388750 19.21500  1
        am     gear    carb
1 1.0000000 5.000000 2.000000
2 1.0000000 4.333333 4.666667
3 0.1428571 3.285714 3.500000
4 0.7000000 4.000000 1.500000
5 0.0000000 3.500000 2.500000
> detach(mtcars)


== Circular plot ==
# Another example: select rows with a minimum value from a certain column (yval in this case)
* http://freakonometrics.hypotheses.org/20667 which uses https://cran.r-project.org/web/packages/circlize/ circlize] package.
> mydf <- read.table(header=T, text='
* https://www.biostars.org/p/17728/
id xval yval
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
A 1  1
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.
A -2  2
B 3  3
B 4  4
C 5  5
')
> x = mydf$xval
> y = mydf$yval
> aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
  id xval yval
1  A    1    1
2  B    3    3
3  C    5    5
</pre>
</li>
</ul>


== Word cloud ==
=== Mean by Group ===
* [http://www.sthda.com/english/wiki/text-mining-and-word-cloud-fundamentals-in-r-5-simple-steps-you-should-know Text mining and word cloud fundamentals in R : 5 simple steps you should know]
[https://statisticsglobe.com/mean-by-group-in-r Mean by Group in R (2 Examples) | dplyr Package vs. Base R]
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
<pre>
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                          # Specify function (i.e. mean)
</pre>
<pre>
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                        # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
              list(name = mean))              # Specify function
</pre>
* [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/ave ave(x, ..., FUN)],
* aggregate(x, by, FUN),
* by(x, INDICES, FUN): return is a list
* tapply(): return results as a matrix or array. Useful for [https://en.wikipedia.org/wiki/Jagged_array ragged array].


== World map ==
== Apply family ==
[https://www.enchufa2.es/archives/visualising-ssh-attacks-with-r.html Visualising SSH attacks with R] ([https://cran.r-project.org/package=rworldmap rworldmap] and [https://cran.r-project.org/package=rgeolocate rgeolocate] packages)
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.  


== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
The following list gives a hierarchical relationship among these functions.
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
*** '''replicate'''(n, expr, simplify = "array")
** '''mapply'''(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
*** '''Vectorize'''(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
*** '''Map'''(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
** '''vapply'''(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
** '''rapply'''(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
* '''tapply'''(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a [https://en.wikipedia.org/wiki/Jagged_array "Ragged" Array]. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
** '''aggregate'''(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each '''columns''' of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame. This is used to '''summarize''' data.
** '''by'''(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each '''subset data frame''' split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
* '''eapply'''(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment


=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
[https://www.queryhome.com/tech/76799/r-difference-between-apply-vs-sapply-vs-lapply-vs-tapply Difference between apply vs sapply vs lapply vs tapply?]
* http://rich-iannone.github.io/DiagrammeR/
* apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
* https://donlelek.github.io/2015-03-31-dags-with-r/
* lapply - When you want to apply a function to each element of a list in turn and get a list back.
* sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
* tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.


=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
Some short examples:
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].  
* [http://www.datasciencemadesimple.com/apply-function-r/ Apply Function in R – apply vs lapply vs sapply vs mapply vs tapply vs rapply vs vapply] from datasciencemadesimple.com.
* [https://stackoverflow.com/a/7141669 How to use which one (apply family) when?]


=== DAGitty (browser-based and R package) ===
=== Apply vs for loop ===
* http://dagitty.net/
Note that, apply's performance is not always better than a for loop. See
* https://cran.r-project.org/web/packages/dagitty/index.html
* http://tolstoy.newcastle.edu.au/R/help/06/05/27255.html (answered by Brian Ripley)
* https://stat.ethz.ch/pipermail/r-help/2014-October/422455.html (has one example)
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]. The author said '' 'an important reason for using *apply() functions may instead be that they fit the functional programming paradigm better, where everything is done using function calls and side effects are reduced'... The scope of the variables defined within f is limited to f, and variables defined outside f cannot be modified inside f (except using the special scoping assignment operator <<-).  ''
** [http://adv-r.had.co.nz/Functional-programming.html Functional programming]
* [https://privefl.github.io/blog/why-loops-are-slow-in-r/ Why loops are slow in R]
* [https://stackoverflow.com/a/18763102 Why is `unlist(lapply)` faster than `sapply`?]


=== dagR ===
=== Progress bar ===
* https://cran.r-project.org/web/packages/dagR
[http://peter.solymos.org/code/2016/09/11/what-is-the-cost-of-a-progress-bar-in-r.html What is the cost of a progress bar in R?]


== Venn Diagram ==
The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out [http://www.theanalystatlarge.com/for-loop-tracking-windows-progress-bar/ this post]. It uses  winProgressBar() and setWinProgressBar() functions.
* limma http://www.ats.ucla.edu/stat/r/faq/venn.htm - only black and white?
* VennDiagram - input has to be the numbers instead of the original vector?
* http://manuals.bioinformatics.ucr.edu/home/R_BioCondManual#TOC-Venn-Diagrams and the [http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R R code] or the [http://www.bioconductor.org/packages/release/bioc/html/systemPipeR.html Bioc package systemPipeR]
<syntaxhighlight lang='rsplus'>
# systemPipeR package method
library(systemPipeR)
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
OLlist <- overLapper(setlist[1:3], type="vennsets")
vennPlot(list(OLlist))                           


# R script source method
[https://www.jottr.org/2020/07/04/progressr-erum2020-slides/ e-Rum 2020 Slides on Progressr] by Henrik Bengtsson. [https://www.jottr.org/2021/06/11/progressr-0.8.0/ progressr 0.8.0: RStudio's progress bar, Shiny progress updates, and absolute progress], [https://www.r-bloggers.com/2022/06/progressr-0-10-1-plyr-now-supports-progress-updates-also-in-parallel/ progressr 0.10.1: Plyr Now Supports Progress Updates also in Parallel]
source("http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R")
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
# or (obtained by dput(setlist))
setlist <- structure(list(A = c("o", "h", "u", "p", "i", "s", "a", "w",  
"b", "z", "n", "c", "k", "j", "y", "m", "t", "q"), B = c("h",
"r", "x", "y", "b", "t", "d", "o", "m", "q", "g", "v", "c", "u",
"f", "z"), C = c("b", "e", "t", "u", "s", "j", "o", "k", "d",
"l", "g", "i", "w", "n", "p", "a", "y", "x", "m", "z"), D = c("f",
"g", "b", "k", "j", "m", "e", "q", "i", "d", "o", "l", "c", "t",
"x", "r", "s", "u", "w", "a", "z", "n"), E = c("u", "w", "o",
"k", "n", "h", "p", "z", "l", "m", "r", "d", "q", "s", "x", "b",
"v", "t"), F = c("o", "j", "r", "c", "l", "l", "u", "b", "f",
"d", "u", "m", "y", "t", "y", "s", "a", "g", "t", "m", "x", "m"
)), .Names = c("A", "B", "C", "D", "E", "F"))


OLlist <- overLapper(setlist[1:3], type="vennsets")
=== simplify option in sapply() ===
counts <- list(sapply(OLlist$Venn_List, length)
<pre>
vennPlot(counts=counts)                         
library(KEGGREST)
</syntaxhighlight>


[[File:Vennplot.png|250px]]
names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2


== Bump chart/Metro map ==
names3 <- keggGet(c("hsa05340"))
https://dominikkoch.github.io/Bump-Chart/
names4 <- sapply(names3, function(x) x$GENE)
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify
# [1] 1
</pre>


== Amazing plots ==
=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
=== New R logo 2/11/2016 ===
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
* http://rud.is/b/2016/02/11/plot-the-new-svg-r-logo-with-ggplot2/
<ul>
* https://www.stat.auckland.ac.nz/~paul/Reports/Rlogo/Rlogo.html
<li>mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
<syntaxhighlight lang='rsplus'>
<pre>
library(sp)
mapply(rep, 1:4, 4:1)
library(maptools)
mapply(rep, times = 1:4, x = 4:1)
library(ggplot2)
mapply(function(x, y) seq_len(x) + y,
library(ggthemes)
      c(a =  1, b = 2, c = 3)# names from first
# rgeos requires the installation of GEOS from http://trac.osgeo.org/geos/
      c(A = 10, B = 0, C = -10))
system("curl http://download.osgeo.org/geos/geos-3.5.0.tar.bz2 | tar jx")
mapply(c, firstList, secondList, SIMPLIFY=FALSE)
system("cd geos-3.5.0; ./configure; make; sudo make install")
</pre>
library(rgeos)
</li>
<li>[https://bensstats.wordpress.com/2020/10/06/robservations-3-finding-the-expected-value-of-the-maximum-of-two-bivariate-normal-variables-with-simulation/ Finding the Expected value of the maximum of two Bivariate Normal variables with simulation] sapply + mapply.
r_wkt_gist_file <- "https://gist.githubusercontent.com/hrbrmstr/07d0ccf14c2ff109f55a/raw/db274a39b8f024468f8550d7aeaabb83c576f7ef/rlogo.wkt"
<pre>
if (!file.exists("rlogo.wkt")) download.file(r_wkt_gist_file, "rlogo.wkt")
z <- mapply(function(u, v) { max(u, v) },
rlogo <- readWKT(paste0(readLines("rlogo.wkt", warn=FALSE))) # rgeos
            u = x[, 1], v = x[, 2])
</pre>
rlogo_shp <- SpatialPolygonsDataFrame(rlogo, data.frame(poly=c("halo", "r"))) # sp
</li>
rlogo_poly <- fortify(rlogo_shp, region="poly") # ggplot2
<li>[http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming </li>
<li>Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
ggplot(rlogo_poly) +
<ul>
  geom_polygon(aes(x=long, y=lat, group=id, fill=id)) +
<li>If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply().
  scale_fill_manual(values=c(halo="#b8babf", r="#1e63b5")) +
{{Pre}}
  coord_equal() +
# Syntax: Map(f, ...)
  theme_map() +
  theme(legend.position="none")
</syntaxhighlight>


=== 3D plot ===
xs <- replicate(5, runif(10), simplify = FALSE)
Using [https://chitchatr.wordpress.com/2010/06/28/fun-with-persp-function/ persp] function to create the following plot. Code in [https://gist.github.com/arraytools/ef955d017cb6b9ef0690a4fe79f809f9 github].
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


[[File:3dpersp.png|200px]]
# instead of a more clumsy way
lapply(seq_along(xs), function(i) {
  weighted.mean(xs[[i]], ws[[i]])
})
</pre>
</li>
<li>Reduce() reduces a vector, x, to a single value by <span style="color: red">recursively</span> calling a function, f, two arguments at a time. A good example of using '''Reduce()''' function is to read a list of matrix files and merge them. See [https://stackoverflow.com/questions/29820029/how-to-combine-multiple-matrix-frames-into-one-using-r How to combine multiple matrix frames into one using R?]  
{{Pre}}
# Syntax: Reduce(f, x, ...)


=== Christmas tree ===
> m1 <- data.frame(id=letters[1:4], val=1:4)
http://wiekvoet.blogspot.com/2014/12/merry-christmas.html. Code in [https://gist.github.com/arraytools/668404a33d32a6652d4dddf5d294689e github].
> m2 <- data.frame(id=letters[2:6], val=2:6)
 
> merge(m1, m2, "id", all = T)
[[File:XMastree.png|150px]]
  id val.x val.y
 
1  a    1    NA
=== Happy Thanksgiving ===
2  b    2    2
[http://blog.revolutionanalytics.com/2015/11/happy-thanksgiving.html Turkey]
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
> m <- list(m1, m2)
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
  id val.x val.y
1  a    1    NA
2  b    2    2
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
</pre>
</li>
</ul>
</li>
</ul>
* [https://statcompute.wordpress.com/2018/09/08/playing-map-and-reduce-in-r-subsetting/ Playing Map() and Reduce() in R – Subsetting] - using parallel and future packages. [https://statcompute.wordpress.com/2018/09/22/union-multiple-data-frames-with-different-column-names/ Union Multiple Data.Frames with Different Column Names]


[[File:Turkey.png|150px]]
=== sapply & vapply ===
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
* sapply can be used in plotting; for example, [https://cran.r-project.org/web/packages/glmnet/vignettes/relax.pdf#page=13 glmnet relax vignette] uses '''sapply(myList, lines, col="grey") ''' to draw multiple lines simultaneously on a list of matrices.


=== Happy Valentine's Day ===
See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
* [https://rud.is/b/2017/02/14/geom%E2%9D%A4%EF%B8%8F/  Geom❤️] 2017
* [http://www.theanalyticslab.nl/2019/02/14/nerds-on-valentines-day/ Happy Valentines day by Nerds] 2019


=== treemap ===
=== rapply - recursive version of lapply ===
http://ipub.com/treemap/
* http://4dpiecharts.com/tag/recursive/
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


[[File:TreemapPop.png|150px]]
=== replicate ===
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
{{Pre}}
> replicate(5, rnorm(3))
          [,1]      [,2]      [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,] 0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</pre>


=== [https://en.wikipedia.org/wiki/Voronoi_diagram Voronoi diagram] ===
See [[#parallel_package|parSapply()]] for a parallel version of replicate().
* https://www.stat.auckland.ac.nz/~paul/Reports/VoronoiTreemap/voronoiTreeMap.html
* http://letstalkdata.com/2014/05/creating-voronoi-diagrams-with-ggplot/


=== Silent Night ===
=== Vectorize ===
[[File:Silentnight.png|200px]]
* [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/Vectorize Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE)]: creates a function wrapper that vectorizes a scalar function. Its value is a list or vector or array. It calls '''mapply()'''.
{{Pre}}
> rep(1:4, 4:1)
[1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
[[1]]
[1] 1 1 1 1


The code in [https://gist.github.com/arraytools/6d841de27bec48fa0b72559e9aeb13ad github].
[[2]]
[1] 2 2 2


=== The Travelling Salesman Portrait ===
[[3]]
https://fronkonstin.com/2018/04/04/the-travelling-salesman-portrait/
[1] 3 3


=== Moon phase calendar ===
[[4]]
https://chichacha.netlify.com/2018/05/26/making-calendar-with-ggplot-moon-phase-calendar/
[1] 4
</pre>
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy]
{{Pre}}
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
[1] 1.6491761 0.9610109
</pre>
* https://blogs.msdn.microsoft.com/gpalem/2013/03/28/make-vectorize-your-friend-in-r/ 
{{Pre}}
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15


=== Calendar heatmap ===
myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
https://stackoverflow.com/questions/26171068/add-dates-to-calendar-heat-map-r
myfunc2(1, 2) # 2
myfunc2(3, 5) # 15
myfunc2(c(1,3), c(2,5)) # NA
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used
</pre>


=== Chaos ===
== plyr and dplyr packages ==
[https://fronkonstin.com/2019/01/10/rcpp-camaron-de-la-isla-and-the-beauty-of-maths/ Rcpp, Camarón de la Isla and the Beauty of Maths]
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]


== Google Analytics ==
[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
=== GAR package ===
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html


== Linear Programming ==
[http://seananderson.ca/courses/12-plyr/plyr_2012.pdf A quick introduction to plyr] with a summary of apply functions in R and compare them with functions in plyr package.
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/


== Linear Algebra ==
# plyr has a common syntax -- easier to remember
* [https://jimskinner.github.io/post/elegant-linear-algebra-in-r-with-the-matrix-package/ Elegant linear algebra in R with the Matrix package]. Matrix package is used.
# plyr requires less code since it takes care of the input and output format
* [https://datascienceplus.com/linear-algebra-for-machine-learning-and-deep-learning-in-r/ Linear Algebra for Machine Learning and Deep Learning in R]. MASS library is used.
# plyr can easily be run in parallel -- faster


== Amazon Alexa ==
Tutorials
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.


== R and Singularity ==
Examples of using dplyr:
https://www.rstudio.com/rviews/2017/03/29/r-and-singularity/
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.


== Teach kids about R with Minecraft ==
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.


== Secure API keys ==
Tibble objects
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
* it does not have row names (cf data frame),
* it never changes the type of the inputs (e.g. it never converts strings to factors!),
* it never changes the names of variables


== Hide a password: keyring package ==
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
* https://cran.r-project.org/web/packages/keyring/index.html
* [http://theautomatic.net/2019/06/25/how-to-hide-a-password-in-r-with-the-keyring-package/ How to hide a password in R with the Keyring package]


== Vision and image recognition ==
{{Pre}}
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
> data(pew, package = "efficient")
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image
> dim(pew)
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"    "tbl"        "data.frame"


== Turn pictures into coloring pages ==
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
# A tibble: 162 x 3
 
                                                      religion Income Count
== Numerical optimization ==
                                                          <chr>  <chr> <int>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/uniroot.html uniroot]: One Dimensional Root (Zero) Finding. This is used in [http://onlinelibrary.wiley.com/doi/10.1002/sim.7178/full simulating survival data for predefined censoring rate]
1                                                    Agnostic  <$10k    27
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
2                                                      Atheist  <$10k    12
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.
...
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
[1] NA
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count, :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
[1] 181.6975
</pre>


= R packages =
To show all rows of a tibble object, use the '''print()''' method.
[[R_packages|R packages]]
<pre>
print(tbObj, n= Inf)


= Tricks =
tbObj %>% print(n= nrow(.))
</pre>


== Getting help ==
If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
* https://stat.ethz.ch/pipermail/r-help/
* https://stat.ethz.ch/pipermail/r-devel/


== Better Coder/coding, best practices ==
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
{{Pre}}
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
TibbleObject$VarName
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
# OR
TibbleObject[["VarName"]]
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore


== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
# OR
dplyr::select(TibbleObject, 2:5) #
</pre>


== Change default R repository ==
'''Convert a data frame to a tibble''' See [http://www.sthda.com/english/wiki/tibble-data-format-in-r-best-and-modern-way-to-work-with-your-data Tibble Data Format in R: Best and Modern Way to Work with Your Data]
[[Rstudio#Change_repository|Change R repository]]
<pre>
my_data <- as_tibble(iris)
class(my_data)
</pre>


Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local .Rprofile settings take precedence.
To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)


For example, I can specify the R mirror I like by creating a single line <.Rprofile> file under my home directory.
=== llply() ===
llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.
<pre>
<pre>
local({
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
  r = getOption("repos")
</pre>
  r["CRAN"] = "https://cran.rstudio.com/"
where rLLID is a list of entrez ID. For example,
  options(repos = r)
<pre>
})
get("org.Hs.egGO")[["6772"]]
options(continue = "  ")
</pre>
message("Hi MC, loading ~/.Rprofile")
returns a list of 49 GOs.
if (interactive()) {
 
  .Last <- function() try(savehistory("~/.Rhistory"))
=== ddply() ===
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
 
=== ldply() ===
[http://rpsychologist.com/an-r-script-to-automatically-look-at-pubmed-citation-counts-by-year-of-publication/ An R Script to Automatically download PubMed Citation Counts By Year of Publication]
 
=== Performance/speed comparison ===
[https://www.r-bloggers.com/2023/01/performance-comparison-of-converting-list-to-data-frame-with-r-language/ Performance comparison of converting list to data.frame with R language]
 
== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
 
=== get_seed() ===
See the same blog
{{Pre}}
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
}
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== Random seeds ===
By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See [https://stat.ethz.ch/R-manual/R-patched/library/base/html/Random.html ?Random].
<pre>
set.seed(as.numeric(Sys.time()))
set.seed(as.numeric(Sys.Date()))  # same seed for each day
</pre>
</pre>


== Change the default web browser ==
=== .Machine and the largest integer, double ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey).
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].
<syntaxhighlight lang='rsplus'>
{{Pre}}
> help.start()
                          Linux/Mac  32-bit Windows 64-bit Windows
If the browser launched by 'xdg-open' is already running, it is *not*
double.eps              2.220446e-16  2.220446e-16  2.220446e-16
    restarted, and you must switch to its window.
double.neg.eps          1.110223e-16  1.110223e-16  1.110223e-16
Otherwise, be patient ...
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
> /usr/bin/xdg-open: 461: /usr/bin/xdg-open: x-www-browser: not found
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: firefox: not found
double.base            2.000000e+00  2.000000e+00  2.000000e+00
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: mozilla: not found
double.digits          5.300000e+01  5.300000e+01  5.300000e+01
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: epiphany: not found
double.rounding        5.000000e+00  5.000000e+00  5.000000e+00
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: konqueror: not found
double.guard            0.000000e+00  0.000000e+00  0.000000e+00
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: chromium-browser: not found
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: google-chrome: not found
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links2: not found
double.exponent        1.100000e+01  1.100000e+01  1.100000e+01
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links: not found
double.min.exp        -1.022000e+03  -1.022000e+03  -1.022000e+03
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: lynx: not found
double.max.exp          1.024000e+03  1.024000e+03  1.024000e+03
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: w3m: not found
integer.max            2.147484e+09  2.147484e+09  2.147484e+09
xdg-open: no method available for opening 'http://127.0.0.1:27919/doc/html/index.html'
sizeof.long            8.000000e+00  4.000000e+00  4.000000e+00
</syntaxhighlight>
sizeof.longlong        8.000000e+00  8.000000e+00  8.000000e+00
sizeof.longdouble      1.600000e+01  1.200000e+01  1.600000e+01
sizeof.pointer          8.000000e+00  4.000000e+00  8.000000e+00
</pre>


The solution is to put
=== NA when overflow ===
<pre>
<pre>
options(browser='seamonkey')
tmp <- 156287L
tmp*tmp
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647
</pre>
</pre>
in the '''.Rprofile''' of your home directory. If the browser is not in the global PATH, we need to put the full path above.


For one-time only purpose, we can use the ''browser'' option in help.start() function:
== How to select a seed for simulation or randomization ==
<syntaxhighlight lang='rsplus'>
* [https://sciprincess.wordpress.com/2019/03/14/how-to-select-a-seed-for-simulation-or-randomization/ How to select a seed for simulation or randomization]
> help.start(browser="seamonkey")
* [https://www.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]
If the browser launched by 'seamonkey' is already running, it is *not*
 
    restarted, and you must switch to its window.
== set.seed() allow alphanumeric seeds ==
Otherwise, be patient ...
https://stackoverflow.com/a/10913336
</syntaxhighlight>
 
 
== set.seed(), for loop and saving random seeds ==
We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
<ul>
* [https://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
<li>[https://www.jottr.org/2020/09/21/detect-when-the-random-number-generator-was-used/ Detect When the Random Number Generator Was Used]
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html
 
== Getting user's home directory ==
See [https://cran.r-project.org/bin/windows/base/rw-FAQ.html#What-are-HOME-and-working-directories_003f What are HOME and working directories?]
<syntaxhighlight lang='rsplus'>
# Windows
normalizePath("~")  # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")  # "C:/Users/brb/Documents"
 
# Mac
normalizePath("~")  # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # "/Users/brb"
 
# Linux
normalizePath("~")  # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # [1] "/home/brb"
</syntaxhighlight>
 
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
* [http://blog.revolutionanalytics.com/2015/11/how-to-store-and-use-authentication-details-with-r.html How to store and use webservice keys and authentication details]
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]
 
If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
<pre>
<pre>
R_LIBS_SITE=F:/R/library
if (interactive()) {
  invisible(addTaskCallback(local({
    last <- .GlobalEnv$.Random.seed
   
    function(...) {
      curr <- .GlobalEnv$.Random.seed
      if (!identical(curr, last)) {
        msg <- "NOTE: .Random.seed changed"
        if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
        message(msg)
        last <<- curr
      }
      TRUE
    }
  }), name = "RNG tracker"))
}
</pre>
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''.
</li>
 
<li>http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
Note that on Windows OS, R/etc contains
<pre>
<pre>
$ ls -l /c/Progra~1/r/r-3.2.0/etc
set.seed(1001)
total 142
data <- vector("list", 30)
-rw-r--r--    1  Administ    1043 Jun 20  2013 Rcmd_environ
seeds <- vector("list", 30)
-rw-r--r--    1   Administ    1924 Mar 17  2010 Rconsole
for(i in 1:30) {
-rw-r--r--    1   Administ      943 Oct  3  2011 Rdevga
   seeds[[i]] <- .Random.seed
-rw-r--r--    1  Administ      589 May 20  2013 Rprofile.site
   data[[i]] <- runif(5)
-rw-r--r--    1  Administ   251894 Jan 17  2015 curl-ca-bundle.crt
}
drwxr-xr-x    1  Administ        0 Jun  8 10:30 i386
   
-rw-r--r--    1  Administ    1160 Dec 31 2014 repositories
# If we save and load .Random.seed from a file using scan(), make
-rw-r--r--    1  Administ    30188 Mar 17  2010 rgb.txt
# sure to convert its type from doubles to integers.
drwxr-xr-x    3  Administ        0 Jun  8 10:30 x64
# Otherwise, .Random.seed will complain!
 
$ ls /c/Progra~1/r/r-3.2.0/etc/i386
Makeconf


$ cat /c/Progra~1/r/r-3.2.0/etc/Rconsole
.Random.seed <- seeds[[23]]  # restore
# Optional parameters for the console and the pager
data.23 <- runif(5)
# The system-wide copy is in R_HOME/etc.
data.23
# A user copy can be installed in `R_USER'.
data[[23]]
</pre>
</li>
</ul>
* [https://www.rdocumentation.org/packages/impute/versions/1.46.0/topics/impute.knn impute.knn]
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].


## Style
== sample() ==
# This can be `yes' (for MDI) or `no' (for SDI).
=== sample() inaccurate on very large populations, fixed in R 3.6.0 ===
  MDI = yes
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17494 The default method for generating from a discrete uniform distribution (used in ‘sample()’, for instance) has been changed]. In prior versions, the probability of generating each integer could vary from equal by up to 0.04% (or possibly more if generating more than a million different integers). See also [https://www.r-bloggers.com/whats-new-in-r-3-6-0/amp/ What's new in R 3.6.0] by David Smith.
# MDI = no
{{Pre}}
# R 3.5.3
set.seed(123)
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1
# 400070 599930
</pre>
* [https://blog.daqana.com/en/fast-sampling-support-in-dqrng/ Fast sampling support in dqrng]
* Differences of the output of sample()
{{Pre}}
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5


# the next two are only relevant for MDI
# R 3.6.0
toolbar = yes
# docker run --net=host -it --rm r-base:3.6.0
statusbar = no
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5
</pre>


## Font.
=== Getting different results with set.seed() in RStudio ===
# Please use only fixed width font.
[https://community.rstudio.com/t/getting-different-results-with-set-seed/31624/2 Getting different results with set.seed()].  ''It's possible that you're loading an R package that is changing the requested random number generator; RNGkind().''
# If font=FixedFont the system fixed font is used; in this case
# points and style are ignored. If font begins with "TT ", only
# True Type fonts are searched for.
font = TT Courier New
points = 10
style = normal # Style can be normal, bold, italic


# Dimensions (in characters) of the console.
=== dplyr::sample_n() ===
rows = 25
The function has a parameter [https://dplyr.tidyverse.org/reference/sample.html weight]. For example if we have some download statistics for each day and we want to do sampling based on their download numbers, we can use this function.
columns = 80
# Dimensions (in characters) of the internal pager.
pgrows = 25
pgcolumns = 80
# should options(width=) be set to the console width?
setwidthonresize = yes


# memory limits for the console scrolling buffer, in chars and lines
== Regular Expression ==
# NB: bufbytes is in bytes for R < 2.7.0, chars thereafter.
See [[Regular_expression|here]].
bufbytes = 250000
buflines = 8000


# Initial position of the console (pixels, relative to the workspace for MDI)
== Read rrd file ==
# xconsole = 0
* https://en.wikipedia.org/wiki/RRDtool
# yconsole = 0
* http://oss.oetiker.ch/rrdtool/
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/


# Dimension of MDI frame in pixels
== on.exit() ==
# Format (w*h+xorg+yorg) or use -ve w and h for offsets from right bottom
Examples of using on.exit(). In all these examples, '''add = TRUE''' is used in the on.exit() call to ensure that each exit action is added to the list of actions to be performed when the function exits, rather than replacing the previous actions.
# This will come up maximized if w==0
<ul>
# MDIsize = 0*0+0+0
<li>Database connections
# MDIsize = 1000*800+100+0
<pre>
# MDIsize = -50*-50+50+50  # 50 pixels space all round
library(RSQLite)
 
sqlite_get_query <- function(db, sql) {
# The internal pager can displays help in a single window
  conn <- dbConnect(RSQLite::SQLite(), db)
# or in multiple windows (one for each topic)
  on.exit(dbDisconnect(conn), add = TRUE)
# pagerstyle can be set to `singlewindow' or `multiplewindows'
  dbGetQuery(conn, sql)
pagerstyle = multiplewindows
}
 
</pre>
## Colours for console and pager(s)
<li>File connections
# (see rwxxxx/etc/rgb.txt for the known colours).
<pre>
background = White
read_chars <- function(file_name) {
normaltext = NavyBlue
  conn <- file(file_name, "r")
usertext = Red
  on.exit(close(conn), add = TRUE)
highlight = DarkRed
  readChar(conn, file.info(file_name)$size)
 
}
## Initial position of the graphics window
</pre>
## (pixels, <0 values from opposite edge)
<li>Temporary files
xgraphics = -25
<pre>
ygraphics = 0
history_lines <- function() {
 
  f <- tempfile()
## Language for messages
  on.exit(unlink(f), add = TRUE)
language =
  savehistory(f)
 
  readLines(f, encoding = "UTF-8")
## Default setting for console buffering: 'yes' or 'no'
}
buffered = yes
</pre>
</pre>
and on Linux
<li>Printing messages
<pre>
<pre>
brb@brb-T3500:~$ whereis R
myfun = function(x) {
R: /usr/bin/R /etc/R /usr/lib/R /usr/bin/X11/R /usr/local/lib/R /usr/share/R /usr/share/man/man1/R.1.gz
  on.exit(print("first"))
  on.exit(print("second"), add = TRUE)
  return(x)
}
</pre>
</ul>


brb@brb-T3500:~$ ls /usr/lib/R
== file, connection ==
bin  COPYING  etc  lib  library  modules  site-library  SVN-REVISION
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
* read() and write()
* read.table() and write.table()
{{Pre}}
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)


brb@brb-T3500:~$ ls /usr/lib/R/etc
foo <- function() {
javaconf  ldpaths  Makeconf  Renviron  Renviron.orig  Renviron.site  Renviron.ucf  repositories  Rprofile.site
  con <- file()
 
  ...
brb@brb-T3500:~$ ls /usr/local/lib/R
  on.exit(close(con))
site-library
  ...
}
</pre>
</pre>
and
[https://r.789695.n4.nabble.com/Why-I-get-this-error-Error-in-close-connection-f-invalid-connection-td904413.html Error in close.connection(f) : invalid connection]. If we want to use '''close(con)''', we have to specify how to '''open''' the connection; such as
<pre>
<pre>
brb@brb-T3500:~$ cat /usr/lib/R/etc/Rprofile.site
con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
##                                              Emacs please make this -*- R -*-
x <- read.delim(con)
## empty Rprofile.site for R on Debian
close(x)
##
</pre>
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Rprofile and Rprofile.site


# ## Example of .Rprofile
=== withr package ===
# options(width=65, digits=5)
https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].
# options(show.signif.stars=FALSE)
# setHook(packageEvent("grDevices", "onLoad"),
#        function(...) grDevices::ps.options(horizontal=FALSE))
# set.seed(1234)
# .First <- function() cat("\n  Welcome to R!\n\n")
# .Last <- function()  cat("\n  Goodbye!\n\n")


# ## Example of Rprofile.site
== Clipboard (?connections), textConnection(), pipe() ==
# local({
<ul>
#  # add MASS to the default packages, set a CRAN mirror
<li>On Windows, we can use readClipboard() and writeClipboard().  
#  old <- getOption("defaultPackages"); r <- getOption("repos")
{{Pre}}
#  r["CRAN"] <- "http://my.local.cran"
source("clipboard")
# options(defaultPackages = c(old, "MASS"), repos = r)
read.table("clipboard")
#})
</pre></li>
brb@brb-T3500:~$ cat /usr/lib/R/etc/Renviron.site
<li>Clipboard -> R. Reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function:
##                                              Emacs please make this -*- R -*-
{{Pre}}
## empty Renviron.site for R on Debian
x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
##
# Or on Mac
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
x <- read.delim(pipe("pbpaste"))
##
# safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
## see help(Startup) for documentation on ~/.Renviron and Renviron.site
</pre>
 
An example is to copy data from [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others?answertab=active#tab-top this post]. In this case we need to use read.table() instead of read.delim().
# ## Example ~/.Renviron on Unix
</li>
# R_LIBS=~/R/library
<li>R -> clipboard on Mac. Note: '''pbcopy''' and '''pbpaste''' are macOS terminal commands. See [http://osxdaily.com/2007/03/05/manipulating-the-clipboard-from-the-command-line/ pbcopy & pbpaste: Manipulating the Clipboard from the Command Line].
# PAGER=/usr/local/bin/less
* pbcopy: takes standard input and places it in the clipboard buffer
 
* pbpaste: takes data from the clipboard buffer and writes it to the standard output
# ## Example .Renviron on Windows
{{Pre}}
# R_LIBS=C:/R/library
clip <- pipe("pbcopy", "w")
# MY_TCLTK="c:/Program Files/Tcl/bin"
write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
 
# write.table(data.frame(Var1, Var2), file = clip, row.names=F, quote=F, sep="\t")
# ## Example of setting R_DEFAULT_PACKAGES (from R CMD check)
close(clip)
# R_DEFAULT_PACKAGES='utils,grDevices,graphics,stats'
</pre>
# # this loads the packages in the order given, so they appear on
<li>
# # the search path in reverse order.
<li>Clipboard -> Excel.  
brb@brb-T3500:~$
* Method 1: Paste icon -> Text import wizard -> Delimit (Tab, uncheck Space) or Fixed width depending on the situation -> Finish.
* Method 2: Ctrl+v first. Then choose Data -> Text to Columns. Fixed width -> Next -> Next -> Finish.
</li>
<li>On Linux, we need to install "xclip". See [https://stackoverflow.com/questions/45799496/r-copy-from-clipboard-in-ubuntu-linux R Copy from Clipboard in Ubuntu Linux]. It seems to work.
{{Pre}}
# sudo apt-get install xclip
read.table(pipe("xclip -selection clipboard -o",open="r"))
</pre>
</pre>
</li>
</ul>


=== What is the best place to save Rconsole on Windows platform ===
=== clipr ===
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
[https://cran.rstudio.com/web/packages/clipr/ clipr]: Read and Write from the System Clipboard


My preferred settings:
== read/manipulate binary data ==
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* x <- readBin(fn, raw(), file.info(fn)$size)
* Size: 12
* rawToChar(x[1:16])
* background: black
* See Biostrings C API
* normaltext: white
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow


and others (default options)
== String Manipulation ==
* pagebg: white
* [https://www.gastonsanchez.com/r4strings/ Handling Strings with R](ebook) by Gaston Sanchez.
* pagetext: navy
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* highlight: DarkRed
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
* dataeditbg: white
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
* dataedittext: navy (View() function)
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
* dataedituser: red
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
* editorbg: white (edit() function)
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]
* editortext: black
* [https://statisticaloddsandends.wordpress.com/2022/05/05/a-short-note-on-the-startswith-function/ A short note on the startsWith function]


== Saving and loading history automatically: .Rprofile & local() ==
=== format(): padding with zero ===
* http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html
* '''.Rprofile''' will automatically be loaded when R has started from that directory
* Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
* '''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
* [http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
** You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
** At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
<pre>
options(continue=" ") # default is "+ "
ngenes <- 10
options(editor="nano") # default is "vi" on Linux
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
# options(htmlhelp=TRUE)
# [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
 
</pre>
local((r <- getOption("repos")
 
  r["CRAN"] <- "http://cran.rstudio.com"
=== noquote() ===
  options(repos = r)))
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.
 
.First <- function(){
# library(Hmisc)
cat("\nWelcome at", date(), "\n")
}
 
.Last <- function(){
cat("\nGoodbye at ", date(), "\n")
</pre>  
* https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
* The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
* [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)


'''Linux''' or '''Mac'''
=== stringr package ===
* https://stringr.tidyverse.org/index.html
* [https://stringr.tidyverse.org/articles/from-base.html Vignette compares stringr functions to their base R equivalents]
* When I try to use trimws() on data obtained from readxl::read_excell(), I find trimws() does not work but [https://stringr.tidyverse.org/reference/str_trim.html stringr::str_trim()] works. [https://stackoverflow.com/questions/45050617/trimws-bug-leading-whitespace-not-removed trimws bug? leading whitespace not removed].


In '''~/.profile''' or '''~/.bashrc''' I put:
=== glue package ===
<ul>
<li>[https://cran.r-project.org/web/packages/glue/index.html glue]. Useful in a loop and some function like ggtitle() or ggsave().
<pre>
<pre>
export R_HISTFILE=~/.Rhistory
library(glue)
name <- "Fred"
glue('My name is {name}.')  # My name is Fred.
</pre>
</pre>
In '''~/.Rprofile''' I put:
</li>
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
</ul>
 
=== Raw data type ===
[https://twitter.com/hadleywickham/status/1387747735441395712 Fun with strings], [https://en.wikipedia.org/wiki/Cyrillic_alphabets Cyrillic alphabets]
<pre>
<pre>
if (interactive()) {
a1 <- "А"
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))
a2 <- "A"
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
a1 == a2
}
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41
</pre>
</pre>


'''Windows'''
=== number of characters limit ===
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]
 
== HTTPs connection ==
HTTPS connection becomes default in R 3.2.2. See
* http://blog.rstudio.org/2015/08/17/secure-https-connections-for-r/
* http://blog.revolutionanalytics.com/2015/08/good-advice-for-security-with-r.html


If you launch R by clicking its icon from Windows Desktop, the R starts in '''C:\User\$USER\Documents''' directory. So we can create a new file '''.Rprofile''' in this directory.
[http://developer.r-project.org/blosxom.cgi/R-devel/2016/12/15#n2016-12-15 R 3.3.2 patched] The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)
<pre>
if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>


== R release versions ==
== setInternet2 ==
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.


== Detect number of running R instances in Windows ==
Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
<pre>
<pre>
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
INFO: No tasks are running which match the specified criteria.
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)
</pre>
It seems the bug was fixed in R 3.2-branch. See [https://github.com/wch/r-source/commit/3a02ed3a50ba17d9a093b315bf5f31ffc0e21b89 8/16/2015] patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385098%28v=vs.85%29.aspx InternetOpenUrl()] function of [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385473%28v=vs.85%29.aspx wininet] library. [http://slacksite.com/other/ftp.html This article] and [http://stackoverflow.com/questions/1699145/what-is-the-difference-between-active-and-passive-ftp this post] explain differences of active and passive FTP.
 
The following R command will show the exact svn revision for the R you are currently using.
<pre>
R.Version()$"svn rev"
</pre>


C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
If setInternet2(T), then https protocol is supported in download.file().  


Image Name                    PID Session Name        Session#    Mem Usage
When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.
============================================================================
Rgui.exe                      1096 Console                    1    44,712 K


C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
The setInternet2() function is defined in [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].


Image Name                    PID Session Name        Session#    Mem Usage
'''R up to 3.2.2'''
============================================================================
Rserve.exe                    6108 Console                    1    381,796 K
</pre>
In R, we can use
<pre>
<pre>
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
[1] ""                                                                           
[2] "Image Name                    PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1    44,804 K"
 
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
</pre>
</pre>
See also
* <src/include/Internal.h> (declare do_setInternet2()),
* <src/main/names.c> (show do_setInternet2() in C)
* <src/main/internet.c>  (define do_setInternet2() in C).


== Editor ==
Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file.  If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2,  it is also possible to download from https by setIneternet2(T).
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs


* Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
'''R 3.3.0'''
* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager. The new version (0.98) also provides a new feature for debugging step-by-step. See also [https://www.rstudio.com/rviews/2016/11/11/easy-tricks-you-mightve-missed/ RStudio Tricks]
<pre>
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code. RStudio can also do this (see the bottom of the code panel).
setInternet2 <- function(use = TRUE) {
* [http://rgedit.sourceforge.net/ Rgedit] which includes a feature of splitting screen into two panes and run R in the bottom panel. See [http://www.stattler.com/article/using-gedit-or-rgedit-r here].
    if(!is.na(use)) stop("use != NA is defunct")
* Komodo IDE with browser preview http://www.youtube.com/watch?v=wv89OOw9roI at 4:06 and http://docs.activestate.com/komodo/4.4/editor.html
    NA
}
</pre>


== GUI for Data Analysis ==
Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.


=== Rcmdr ===
== Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan() ==
http://cran.r-project.org/web/packages/Rcmdr/index.html
In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, '''Inf''' and '''NaN''' as input or output.


=== Deducer ===
See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].
http://cran.r-project.org/web/packages/Deducer/index.html


=== jamovi ===
[https://datasciencetut.com/how-to-replace-inf-values-with-na-in-r/ How to replace Inf with NA in All or Specific Columns of the Data Frame]
* https://www.jamovi.org/
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]


== Scope ==
== replace() function ==
See
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
* https://stackoverflow.com/a/11811147
* [[#How_to_exit_a_sourced_R_script|source()]] does not work like C's preprocessor where statements in header files will be literally inserted into the code. It does not work when you define a variable in a function but want to use it outside the function (even through '''source()''')


<syntaxhighlight lang='rsplus'>
== File/path operations ==
## foo.R ##
* list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
cat(ArrayTools, "\n")
* file.info()
## End of foo.R
* dir.create()
* file.create()
* file.copy()
* file.exists()
<ul>
<li>'''basename'''() - remove the parent path, '''dirname'''() - returns the part of the path up to but excluding the last path separator
<pre>
> file.path("~", "Downloads")
[1] "~/Downloads"
> dirname(file.path("~", "Downloads"))
[1] "/home/brb"
> basename(file.path("~", "Downloads"))
[1] "Downloads"
</pre>
</li></ul>
* '''path.expand'''("~/.Renviron")  # "/home/brb/.Renviron"
<ul>
<li> '''normalizePath'''() # Express File Paths in Canonical Form
<pre>
> cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
/usr/lib/R
/tmp/RtmpzvDhAe
</pre>
</li>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/system.file system.file()] - Finds the full file names of files in packages etc
<pre>
> system.file("extdata", "ex1.bam", package="Rsamtools")
[1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
</pre>
</li></ul>
* tools::file_path_sans_ext() - [https://stackoverflow.com/a/29114021 remove the file extension] or the sub() function.


# 1. Error
== read/download/source a file from internet ==
predict <- function() {
=== Simple text file http ===
  ArrayTools <- "C:/Program Files" # or through load() function
<pre>
  source("foo.R")                  # or through a function call; foo()
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
}
</pre>
predict()  # Object ArrayTools not found


# 2. OK. Make the variable global
=== Zip, RData, gz file and url() function ===
predict <- function() {
<pre>
  ArrayTools <<- "C:/Program Files'
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
  source("foo.R")
</pre>
}
<pre>
predict()
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
ArrayTools
source(con)
close(con)
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.
 
Another example is [https://stackoverflow.com/a/9548672 Read gzipped csv directly from a url in R]
<pre>
con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
                      "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))
</pre>
 
Another example of using url() is
<pre>
load(url("http:/www.example.com/example.RData"))
</pre>
 
This does not work with load(), dget(), read.table() for files on '''OneDrive'''. In fact, I cannot use wget with shared files from OneDrive. The following trick works: [https://mangolassi.it/topic/19276/how-to-configure-a-onedrive-file-for-use-with-wget How to configure a OneDrive file for use with wget].


# 3. OK. Create a global variable
'''Dropbox''' is easy and works for load(), wget, ...
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()
</syntaxhighlight>


'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.


Example 1.  
=== zip function ===
This will include 'hallmarkFiles' root folder in the files inside zip.
<pre>
<pre>
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
zip(zipfile = 'myFile.zip',  
> ttt
    files = dir('hallmarkFiles', full.names = TRUE))
  type JpnTest
 
1    a     999
# Verify/view the files. 'list = TRUE' won't extract
2    b    999
unzip('testZip.zip', list = TRUE)  
3    c    999
</pre>
4    d    999
 
5    e    999
=== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ===
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.
> jpntest()
 
  type JpnTest
=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
1    a      N5
{{Pre}}
2    b    999
require(RCurl)
3    c    999
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
4    d    999
read.csv(textConnection(myCsv))
5    e    999
> ttt
  type JpnTest
1    a    999
2    b    999
3    c    999
4    d    999
5    e    999
</pre>
</pre>


Example 2. [http://stackoverflow.com/questions/1236620/global-variables-in-r How can we set global variables inside a function?] The answer is to use the "<<-" operator or '''assign(, , envir = .GlobalEnv)''' function.
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]


Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
=== Github files https using RCurl package ===
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
<pre>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package


Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
== data summary table ==
=== summarytools: create summary tables for vectors and data frames ===
https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.


=== New environment ===
=== skimr: A frictionless, pipeable approach to dealing with summary statistics ===
http://adv-r.had.co.nz/Environments.html
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]


Run the same function on a bunch of R objects
=== modelsummary ===
<syntaxhighlight lang='rsplus'>
[https://cloud.r-project.org/web/packages/modelsummary/index.html modelsummary]: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready
mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(mye[[n]])
</syntaxhighlight>


== Speedup R code ==
=== broom ===
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
[[Tidyverse#broom|Tidyverse->broom]]


== Profiler ==
=== Create publication tables using '''tables''' package ===
(Video) [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler]
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]


== && vs & ==
R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  
{{Pre}}
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                 
                Sepal.Length      Sepal.Width   
Species    n  mean        sd  mean        sd 
setosa      50 5.01        0.35 3.43        0.38
versicolor  50 5.94        0.52 2.77        0.31
virginica  50 6.59        0.64 2.97        0.32
All        150 5.84        0.83 3.06        0.44
> str(iris)
'data.frame':  150 obs. of  5 variables:
$ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
$ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
$ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
$ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
$ Species    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
</pre>
and
<pre>
# This example shows some of the less common options       
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
> z <- rnorm(100)+5
> fmt <- function(x) {
  s <- format(x, digits=2)
  even <- ((1:length(s)) %% 2) == 0
  s[even] <- sprintf("(%s)", s[even])
  s
}
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
                  Status             
Sex    Statistic high  low    medium
Female mean      4.88  4.96  5.17
        sd        (1.20) (0.82) (1.35)
Male  mean      4.45  4.31  5.05
        sd        (1.01) (0.93) (0.75)
</pre>


The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The longer form evaluates left to right examining only the first element of each vector.
=== fgsea example ===
[http://www.bioconductor.org/packages/release/bioc/vignettes/fgsea/inst/doc/fgsea-tutorial.html  vignette] & [https://github.com/ctlab/fgsea/blob/master/R/plot.R#L28 source code]


== stopifnot(): function argument sanity check ==
=== (archived) ClinReport: Statistical Reporting in Clinical Trials ===
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]
https://cran.r-project.org/web/packages/ClinReport/index.html


== Vectorization ==
== Append figures to PDF files ==
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
* [https://github.com/henrikbengtsson/matrixstats matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors). E.g. col / rowMedians(), col / rowRanks(), and col / rowSds(). [https://github.com/HenrikBengtsson/matrixStats/wiki/Benchmark-reports Benchmark reports].


=== sapply vs vectorization ===
== Save base graphics as pseudo-objects ==
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
[https://www.andrewheiss.com/blog/2016/12/08/save-base-graphics-as-pseudo-objects-in-r/ Save base graphics as pseudo-objects in R]. Note there are some cons with this approach.
<pre>
pdf(NULL)
dev.control(displaylist="enable")
plot(df$x, df$y)
text(40, 0, "Random")
text(60, 2, "Text")
lines(stats::lowess(df$x, df$y))
p1.base <- recordPlot()
invisible(dev.off())


=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
# Display the saved plot
split() can be used to split a vector, columns or rows. See [https://stackoverflow.com/questions/3302356/how-to-split-a-data-frame How to split a data frame?]
grid::grid.newpage()
* Split rows of a data frame/matrix <syntaxhighlight lang='rsplus'>
p1.base
split(mtcars,mtcars$cyl)
</pre>
</syntaxhighlight>
* Split columns of a data frame/matrix. <syntaxhighlight lang='rsplus'>
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10)))
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
</syntaxhighlight>
* split() + sapply() to merge columns. See below 'Mean of duplicated columns' for more detail.
* split() + sapply() to split a vector. See [https://www.rdocumentation.org/packages/genefilter/versions/1.54.2/topics/nsFilter nsFilter()] function which can remove duplicated probesets/rows using unique Entrez Gene IDs ('''genefilter''' package). The source code of [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/nsFilter.R#L224 nsFilter()] and [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/all.R#L170 findLargest()]. <syntaxhighlight lang='rsplus'>
tSsp = split.default(testStat, lls)
# testStat is a vector of numerics including probeset IDs as names
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.


sapply(tSsp, function(x) names(which.max(x)))  
== Extracting tables from PDFs ==
# return a vector of probset IDs of length of unique entrez IDs
<ul>
</syntaxhighlight>  
<li>[http://datascienceplus.com/extracting-tables-from-pdfs-in-r-using-the-tabulizer-package/ extracting Tables from PDFs in R] using Tabulizer. This needs the [https://cran.r-project.org/web/packages/rJava/index.html rJava] package. Linux works fine. Some issue came out on my macOS 10.12 Sierra. '''Library not loaded: /Library/Java/JavaVirtualMachines/jdk-9.jdk/Contents/Home/lib/server/libjvm.dylib. Referenced from: /Users/XXXXXXX/Library/R/3.5/library/rJava/libs/rJava.so'''.
: And here is another example from the [https://cran.r-project.org/web/packages/bigmemory/vignettes/Overview.pdf bigmemory] vignette,
</li>
: <syntaxhighlight lang='rsplus'>
<li>
planeindices <- split(1:nrow(x), x[,'TailNum'])
[https://docs.ropensci.org/pdftools/ pdftools] - Text Extraction, Rendering and Converting of PDF Documents. [https://ropensci.org/technotes/2018/12/14/pdftools-20/ pdf_text() and pdf_data()] functions.  
planeStart <- sapply(planeindices,
{{Pre}}
                    function(i) birthmonth(x[i, c('Year','Month'),
library(pdftools)
                                            drop=FALSE]))
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
</syntaxhighlight>
txt <- pdf_text(pdf_file) # length = number of pages
# Suppose the table we are interested in is on page 1
cat(txt[1]) # Good but not in a data frame format
 
pdf_data(pdf_file)[[1]]  # data frame/tibble format
</pre>  
However, it seems it does not work on [http://www.bloodjournal.org/content/109/8/3177/tab-figures-only Table S6]. Tabulizer package is better at this case.
 
This is another example. [https://mp.weixin.qq.com/s?__biz=MzAxMDkxODM1Ng==&mid=2247490327&idx=1&sn=cca7d4423426318e0c23adb098cf0ad7&chksm=9b485bacac3fd2ba2196b380c59b5eab9d29795d3334b040f50a2fa58124ec6e3be9472829e0&scene=21#wechat_redirect 神技能-自动化批量从PDF里面提取表格]
</li>
<li>[https://www.linuxuprising.com/2019/05/how-to-convert-pdf-to-text-on-linux-gui.html?m=1 How To Convert PDF To Text On Linux (GUI And Command Line)]. It works when I tested my PDF file.
{{Pre}}
sudo apt install poppler-utils
pdftotext -layout input.pdf output.txt
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</pre>
</li>
<li>[https://www.adobe.com/acrobat/how-to/pdf-to-excel-xlsx-converter.html Convert PDF files into Excel spreadsheets] using Adobe Acrobat. See [https://helpx.adobe.com/acrobat/how-to/extract-pages-from-pdf.html How to extract pages from a PDF]. Note the PDF file should not be opened by Excel since it is binary format Excel can't recognize.
<li>I found it is easier to use copy the column (it works) from PDF and paste them to Excel </li>
</ul>
 
== Print tables ==
 
=== addmargins() ===
* [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins]. Puts Arbitrary Margins On Multidimensional Tables Or Arrays.
* [https://datasciencetut.com/how-to-put-margins-on-tables-or-arrays-in-r/ How to put margins on tables or arrays in R?]
 
=== tableone ===
* https://cran.r-project.org/web/packages/tableone/
* [https://datascienceplus.com/table-1-and-the-characteristics-of-study-population/ Table 1 and the Characteristics of Study Population]
* [https://www.jianshu.com/p/e76f2b708d45 如何快速绘制论文的表1(基本特征三线表)?]
* See Table 1 from [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]
 
=== Some examples ===
Cox models
* [https://aacrjournals.org/clincancerres/article/27/12/3383/671420/Integrative-Genomic-Analysis-of-Gemcitabine Integrative Genomic Analysis of Gemcitabine Resistance in Pancreatic Cancer by Patient-derived Xenograft Models]
 
=== finalfit package ===
[https://finalfit.org/index.html summary_factorlist()] from the finalfit package.
 
=== table1 ===
* https://cran.r-project.org/web/packages/table1/
* [https://www.rdatagen.net/post/2023-09-26-nice-looking-table-1-with-standardized-mean-difference/ Creating a nice looking Table 1 with standardized mean differences (SMD)]. SMD is the difference in group means divided by the pooled standard deviation (and is defined differently for categorical measures). Note that the pooled standard deviation defined here is different from we see on the '''[[T-test#Two_sample_test_assuming_equal_variance|t.test]]''' when we assume equivalent variance in two samples.
 
=== gtsummary ===
* [https://education.rstudio.com/blog/2020/07/gtsummary/ Presentation-Ready Summary Tables with gtsummary]
* [https://www.danieldsjoberg.com/gtsummary/ gtsummary] & on [https://cloud.r-project.org/web/packages/gtsummary/index.html CRAN]
** [https://www.danieldsjoberg.com/gtsummary/articles/tbl_summary.html tbl_summary()]. The output is in the "Viewer" window.
* An example: [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]. The table is saved in a png file. The column variable is response.
 
=== gt ===
[https://www.r-bloggers.com/2024/02/introduction-to-clinical-tables-with-the-gt-package/ Introduction to Clinical Tables with the {gt} Package]


=== Mean of duplicated columns: rowMeans ===
=== dplyr ===
* [https://stackoverflow.com/questions/35925529/reduce-columns-of-a-matrix-by-a-function-in-r Reduce columns of a matrix by a function in R]  <syntaxhighlight lang='rsplus'>
https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
colnames(x) <- c("A","A", "b", "b", "b", "c"); x
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res


# vapply() is safter than sapply().
=== tables::tabular() ===
# The 3rd arg in vapply() is a template of the return value.
res2 <- vapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
              rep(0, nrow(x)))
</syntaxhighlight>
* [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/colSums colSums, rowSums, colMeans, rowMeans] (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster. <syntaxhighlight lang='rsplus'>
rowMeans(x, na.rm=T)
# [1] 31 27 28 29 30 31 32 33 34 35


apply(x, 1, mean, na.rm=T)
=== gmodels::CrossTable() ===
# [1] 31 27 28 29 30 31 32 33 34 35
https://www.statmethods.net/stats/frequencies.html
</syntaxhighlight>
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)


=== Mean of duplicated rows: colMeans and rowsum ===
=== base::prop.table(x, margin) ===
* colMeans(x, na.rm = FALSE, dims = 1)
[http://developer.r-project.org/blosxom.cgi/R-devel/2020/02/13#n2020-02-13 New function ‘proportions()and ‘marginSums(). These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’.] for R 4.0.0.
: <syntaxhighlight lang='rsplus'>
<pre>
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
R> m <- matrix(1:4, 2)
rownames(x) <- c(rep("a", 2), rep("b", 3), rep("c", 4), "d")
R> prop.table(m, 1) # row percentage
res <- sapply(split(1:nrow(x), rownames(x)),
          [,1]      [,2]
              function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
[1,] 0.2500000 0.7500000
res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
[2,] 0.3333333 0.6666667
</syntaxhighlight>
R> prop.table(m, 2) # column percentage
* rowsum(x, group, reorder = TRUE, …)
          [,1]      [,2]
: <syntaxhighlight lang='rsplus'>
[1,] 0.3333333 0.4285714
x <- matrix(runif(100), ncol = 5) # 20 x 5
[2,] 0.6666667 0.5714286
group <- sample(1:8, 20, TRUE)
</pre>
(xsum <- rowsum(x, group)) # 8 x 5
 
</syntaxhighlight>
=== stats::xtabs() ===
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
* [http://www.statmethods.net/management/aggregate.html aggregate()] function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. [http://www.win-vector.com/blog/2015/10/dont-use-statsaggregate/ Don't use aggregate] post. <syntaxhighlight lang='rsplus'>
> attach(mtcars)
dim(mtcars)
[1] 32 11
> head(mtcars)
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
Mazda RX4        21.0  6  160 110 3.90 2.620 16.46  0  1    4    4
Mazda RX4 Wag    21.0   6  160 110 3.90 2.875 17.02  0 1    4    4
Datsun 710        22.8  4  108  93 3.85 2.320 18.61  1  1    4    1
Hornet 4 Drive    21.4  6  258 110 3.08 3.215 19.44  1  0    3    1
Hornet Sportabout 18.7  8  360 175 3.15 3.440 17.02  0  0    3    2
Valiant          18.1  6  225 105 2.76 3.460 20.22  1  0    3    1
> with(mtcars, table(cyl, vs))
  vs
cyl  0  1
  4  1 10
  6  3  4
  8 14  0
> aggdata <-aggregate(mtcars, by=list(cyl,vs),  FUN=mean, na.rm=TRUE)
> print(aggdata)
  Group.1 Group.2     mpg cyl  disp      hp    drat      wt    qsec vs
1       4      0 26.00000  4 120.30  91.0000 4.430000 2.140000 16.70000  0
2       6      0 20.56667  6 155.00 131.6667 3.806667 2.755000 16.32667  0
3      8      0 15.10000  8 353.10 209.2143 3.229286 3.999214 16.77214  0
4      4      1 26.73000  4 103.62  81.8000 4.035000 2.300300 19.38100  1
5      6      1 19.12500  6 204.55 115.2500 3.420000 3.388750 19.21500  1
        am    gear    carb
1 1.0000000 5.000000 2.000000
2 1.0000000 4.333333 4.666667
3 0.1428571 3.285714 3.500000
4 0.7000000 4.000000 1.500000
5 0.0000000 3.500000 2.500000
> detach(mtcars)


# Another example: select rows with a minimum value from a certain column (yval in this case)
=== stats::ftable() ===
> mydf <- read.table(header=T, text='
{{Pre}}
  id xval yval
> ftable(Titanic, row.vars = 1:3)
  A 1
                  Survived No Yes
  A -2 2
Class Sex    Age                 
  B 3 3
1st  Male  Child            0  5
  B 4 4
            Adult          118 57
  C 5 5
      Female Child            0  1
')
            Adult            4 140
> x = mydf$xval
2nd  Male  Child            0 11
> y = mydf$yval
            Adult          154 14
> aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
      Female Child            0 13
  id xval yval
            Adult          13 80
1 A    1    1
3rd  Male  Child          35 13
2  B    3    3
            Adult          387 75
3 C    5    5
      Female Child          17 14
</syntaxhighlight>
            Adult          89 76
 
Crew Male  Child            0  0
== Apply family ==
            Adult          670 192
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.
      Female Child            0  0
 
            Adult            3  20
The following list gives a hierarchical relationship among these functions.
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
            Survived No Yes
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
Class Sex                   
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
1st  Male            118 62
*** '''replicate'''(n, expr, simplify = "array")
      Female            4 141
** '''mapply'''(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
2nd  Male            154  25
*** '''Vectorize'''(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
      Female          13  93
*** '''Map'''(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
3rd  Male            422  88
** '''vapply'''(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
      Female          106  90
** '''rapply'''(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
Crew  Male            670 192
* '''tapply'''(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a [https://en.wikipedia.org/wiki/Jagged_array "Ragged" Array]. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
      Female            3  20
** '''aggregate'''(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each '''columns''' of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame.
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
** '''by'''(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each '''subset data frame''' split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
            Survived  No Yes
* '''eapply'''(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment
Sex    Class               
 
Male  1st            118  62
[https://www.queryhome.com/tech/76799/r-difference-between-apply-vs-sapply-vs-lapply-vs-tapply Difference between apply vs sapply vs lapply vs tapply?]
      2nd            154  25
* apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
      3rd            422  88
* lapply - When you want to apply a function to each element of a list in turn and get a list back.
      Crew          670 192
* sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
Female 1st              4 141
* tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.
      2nd            13  93
 
      3rd            106  90
Some short examples:
      Crew            3  20
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
> str(Titanic)
* [http://www.datasciencemadesimple.com/apply-function-r/ Apply Function in R – apply vs lapply vs sapply vs mapply vs tapply vs rapply vs vapply] from datasciencemadesimple.com.
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
* [https://stackoverflow.com/a/7141669 How to use which one (apply family) when?]
- attr(*, "dimnames")=List of 4
 
  ..$ Class  : chr [1:4] "1st" "2nd" "3rd" "Crew"
=== Apply vs for loop ===
  ..$ Sex    : chr [1:2] "Male" "Female"
Note that, apply's performance is not always better than a for loop. See
  ..$ Age    : chr [1:2] "Child" "Adult"
* http://tolstoy.newcastle.edu.au/R/help/06/05/27255.html (answered by Brian Ripley)
  ..$ Survived: chr [1:2] "No" "Yes"
* https://stat.ethz.ch/pipermail/r-help/2014-October/422455.html (has one example)
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]
> x
* [https://privefl.github.io/blog/why-loops-are-slow-in-r/ Why loops are slow in R]
          gear  3  4  5
* [https://stackoverflow.com/a/18763102 Why is `unlist(lapply)` faster than `sapply`?]
cyl vs am             
4  0  0        0  0  0
      1        0  0  1
    1  0        1  2  0
      1        0  6  1
6  0  0        0  0  0
      1        0  2  1
    1  0        2  2  0
      1        0  0  0
8  0  0      12  0  0
      1        0  0  2
    1  0        0  0  0
      1        0  0  0
> ftable(x, row.vars = c(2, 4))
        cyl  4    6    8 
        am  0  1  0  1  0  1
vs gear                     
0  3        0  0  0  0 12  0
  4        0  0  0  2  0  0
  5        0  1  0  1  0  2
1  3        1  0  2  0  0  0
  4        2  6  2  0  0  0
  5        0  1  0  0  0  0
>
> ## Start with expressions, use table()'s "dnn" to change labels
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
        dnn = c("Cylinders", "V/S", "Transmission", "Gears"))


=== Progress bar ===
          Cylinders    4    6    8 
[http://peter.solymos.org/code/2016/09/11/what-is-the-cost-of-a-progress-bar-in-r.html What is the cost of a progress bar in R?]
          Transmission  0  1  0  1  0  1
 
V/S Gears                             
The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out [http://www.theanalystatlarge.com/for-loop-tracking-windows-progress-bar/ this post]. It uses  winProgressBar() and setWinProgressBar() functions.
0  3                  0  0  0  0 12  0
    4                  0  0  0  2  0  0
    5                  0  1  0  1  0  2
1  3                  1  0  2  0  0  0
    4                  2  6  2  0  0  0
    5                  0  1  0  0  0  0
</pre>
 
== tracemem, data type, copy ==
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]
 
== Tell if the current R is running in 32-bit or 64-bit mode ==
<pre>
8 * .Machine$sizeof.pointer
</pre>
where '''sizeof.pointer''' returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.


=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
== 32- and 64-bit ==
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
* mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming
* Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
* Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).
** If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply(). <syntaxhighlight lang='rsplus'>
 
# Syntax: Map(f, ...)
== Handling length 2^31 and more in R 3.0.0 ==


xs <- replicate(5, runif(10), simplify = FALSE)
From R News for 3.0.0 release:
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


# instead of a more clumsy way
''There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.
lapply(seq_along(xs), function(i) {
''
  weighted.mean(xs[[i]], ws[[i]])
})
</syntaxhighlight>
** Reduce() reduces a vector, x, to a single value by <span style="color: red">recursively</span> calling a function, f, two arguments at a time. A good example of using '''Reduce()''' function is to read a list of matrix files and merge them. See [https://stackoverflow.com/questions/29820029/how-to-combine-multiple-matrix-frames-into-one-using-r How to combine multiple matrix frames into one using R?] <syntaxhighlight lang='rsplus'>
# Syntax: Reduce(f, x, ...)


> m1 <- data.frame(id=letters[1:4], val=1:4)
In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error
> m2 <- data.frame(id=letters[2:6], val=2:6)
<pre>
> merge(m1, m2, "id", all = T)
> x <- seq(1, 2^31)
  id val.x val.y
Error in from:to : result would be too long a vector
1  a    1    NA
</pre>
2  b    2    2
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
> m <- list(m1, m2)
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
  id val.x val.y
a     1    NA
2  b    2    2
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
</syntaxhighlight>
* [https://statcompute.wordpress.com/2018/09/08/playing-map-and-reduce-in-r-subsetting/ Playing Map() and Reduce() in R – Subsetting] - using parallel and future packages. [https://statcompute.wordpress.com/2018/09/22/union-multiple-data-frames-with-different-column-names/ Union Multiple Data.Frames with Different Column Names]


=== sapply & vapply ===
However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
<pre>
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
> system.time(x <- seq(1,2^31))
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
  user  system elapsed
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
  8.604  11.060 120.815
> length(x)
[1] 2147483648
> length(x)/2^20
[1] 2048
> gc()
            used    (Mb) gc trigger    (Mb)  max used    (Mb)
Ncells    183823    9.9    407500    21.8    350000    18.7
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
>
</pre>
Note:
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].
# [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17330 [[<- for index 2^31 fails]


See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
== NA in index ==
* Question: what is seq(1, 3)[c(1, 2, NA)]?
 
Answer: It will reserve the element with NA in indexing and return the value NA for it.


=== rapply - recursive version of lapply ===
* Question: What is TRUE & NA?
* http://4dpiecharts.com/tag/recursive/
Answer: NA
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


=== replicate ===
* Question: What is FALSE & NA?
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
Answer: FALSE
<syntaxhighlight lang='rsplus'>
> replicate(5, rnorm(3))
          [,1]      [,2]      [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</syntaxhighlight>


See [[#parallel_package|parSapply()]] for a parallel version of replicate().
* Question: c("A", "B", NA) != "" ?
Answer: TRUE TRUE NA


=== Vectorize ===
* Question: which(c("A", "B", NA) != "") ?
* [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/Vectorize Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE)]: creates a function wrapper that vectorizes a scalar function. Its value is a list or vector or array. It calls '''mapply()'''. <syntaxhighlight lang='rsplus'>
Answer: 1 2
> rep(1:4, 4:1)
[1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
[[1]]
[1] 1 1 1 1


[[2]]
* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
[1] 2 2 2
Answer: TRUE TRUE FALSE


[[3]]
* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
[1] 3 3
Answer: TRUE TRUE FALSE


[[4]]
'''Conclusion''': In order to exclude empty or NA for numerical or character data type, we can use '''which()''' or a convenience function '''keep.complete(x) <- function(x) x != "" & !is.na(x)'''. This will guarantee return logical values and not contain NAs.
[1] 4
</syntaxhighlight>
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy] <syntaxhighlight lang='rsplus'>
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
[1] 1.6491761 0.9610109
</syntaxhighlight>
* https://blogs.msdn.microsoft.com/gpalem/2013/03/28/make-vectorize-your-friend-in-r/  <syntaxhighlight lang='rsplus'>
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15


myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
Don't just use x != "" OR !is.na(x).
myfunc2(1, 2) # 2
myfunc2(3, 5) # 15
myfunc2(c(1,3), c(2,5)) # NA
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used
</syntaxhighlight>


== plyr and dplyr packages ==
=== Some functions ===
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
* X %>% [https://tidyr.tidyverse.org/reference/drop_na.html tidyr::drop_na()]
* '''stats::na.omit()''' and '''stats::complete.cases()'''. [https://statisticsglobe.com/na-omit-r-example/ NA Omit in R | 3 Example Codes for na.omit (Data Frame, Vector & by Column)]


[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
== Constant and 'L' ==
Add 'L' after a constant. For example,
{{Pre}}
for(i in 1L:n) { }


[http://seananderson.ca/courses/12-plyr/plyr_2012.pdf A quick introduction to plyr] with a summary of apply functions in R and compare them with functions in plyr package.
if (max.lines > 0L) { }


# plyr has a common syntax -- easier to remember
label <- paste0(n-i+1L, ": ")
# plyr requires less code since it takes care of the input and output format
# plyr can easily be run in parallel -- faster


Tutorials
n <- length(x);  if(n == 0L) { }
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
</pre>
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.


Examples of using dplyr:
== Vector/Arrays ==
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
R indexes arrays from 1 like Fortran, not from 0 like C or Python.
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.  


=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
=== remove integer(0) ===
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
[https://stackoverflow.com/a/27980810 How to remove integer(0) from a vector?]


Tibble objects
=== Append some elements ===
* it does not have row names (cf data frame),
[https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ append() and its after argument]
* it never changes the type of the inputs (e.g. it never converts strings to factors!),  
 
* it never changes the names of variables
=== setNames() ===
Assign names to a vector
 
<pre>
z <- setNames(1:3, c("a", "b", "c"))
# OR
z <- 1:3; names(z) <- c("a", "b", "c")
# OR
z <- c("a"=1, "b"=2, "c"=3) # not work if "a", "b", "c" is like x[1], x[2], x[3].
</pre>
 
== Factor ==
=== labels argument ===
We can specify the factor levels and new labels using the factor() function.


Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
{{Pre}}
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))


<syntaxhighlight lang='rsplus'>
factor(rev(letters[1:3]), labels = c("A", "B", "C"))
> data(pew, package = "efficient")
# C B A
> dim(pew)
# Levels: A B C
[1] 18 10
</pre>
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"     "tbl"       "data.frame"


> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
=== Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when() ===
# A tibble: 162 x 3
* [https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html Mastering Data Segmentation: A Guide to Using the cut() Function in R]
                                                      religion Income Count
:<syntaxhighlight lang='r'>
                                                          <chr>  <chr> <int>
cut(
1                                                    Agnostic  <$10k    27
    c(0, 10, 30),  
2                                                      Atheist  <$10k    12
    breaks = c(0, 30, 50, Inf),  
...
    labels = c("Young", "Middle-aged", "Elderly")
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
) # Default include.lowest = FALSE
[1] NA
# [1] <NA>  Young Young
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count, :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
[1] 181.6975
</syntaxhighlight>
</syntaxhighlight>
* https://dplyr.tidyverse.org/reference/case_when.html
* [https://rpubs.com/DaveRosenman/ifelsealternative Using dplyr’s mutate and case_when functions as alternative for if else statement]
* [http://www.datasciencemadesimple.com/case-statement-r-using-case_when-dplyr/ Case when in R using case_when() Dplyr – case_when in R]
* [https://predictivehacks.com/how-to-convert-continuous-variables-into-categorical-by-creating-bins/ How To Convert Continuous Variables Into Categorical By Creating Bins]
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/cut ?cut]
{{Pre}}
set.seed(1)
x <- rnorm(100)
facVar <- cut(x, c(min(x), -1, 1, max(x)), labels = c("low", "medium", "high"))
table(facVar, useNA = "ifany")
facVar
#  low medium  high  <NA>
#    10    74    15      1
</pre>
Note the option '''include.lowest = TRUE''' is needed when we use cut() + quantile(); otherwise the smallest data will become NA since the intervals have the format '''(a, b]'''.
<pre>
x2 <- cut(x, quantile(x, 0:2/2), include.lowest = TRUE) # split x into 2 levels
x2 <- cut(x, quantile(x, 0:3/3), include.lowest = TRUE) # split x into 3 levels


If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.
library(tidyverse); library(magrittr)
set.seed(1)
breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
x <- runif(50)
bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)


'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
data.frame(sc=x, bins=bins) %>%
<syntaxhighlight lang='rsplus'>
  group_by(bins) %>%
TibbleObject$VarName
  summarise(n=n()) %>%
# OR
  ggplot(aes(x = bins, y = n)) +
TibbleObject[["VarName"]]
    geom_col(color = "black", fill = "#90AACB") +
# OR
    theme_minimal() +
pull(TibbleObject, VarName) # won't be a tibble object anymore
    theme(axis.text.x = element_text(angle = 90)) +
    theme(legend.position = "none") + coord_flip()
</pre>
<li>[https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html A Guide to Using the cut() Function in R]
<li>[https://youtu.be/7oyiPBjLAWY?t=2480 tibble object]
{{Pre}}
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
      age_cat = case_when(
          age_yrs < 2 ~ "baby",
          age_yrs < 13 ~ "kid",
          age_yrs < 20 ~ "teen",
          TRUE        ~ "adult")
)
</pre>
</li>
<li>[https://youtu.be/JsNqXLl3eFc?t=96 R tip: Learn dplyr’s case_when() function]
<pre>
case_when(
  condition1 ~ value1,
  condition2 ~ value2,
  TRUE ~ ValueAnythingElse
)
# Example
case_when(
  x %%2 == 0 ~ "even",
  x %%2 == 1 ~ "odd",
  TRUE ~ "Neither even or odd"
)
</pre>
<li>
</ul>
 
=== How to change one of the level to NA ===
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
<pre>
x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA
</pre>


dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
[https://webbedfeet.netlify.app/post/creating-missing-values-in-factors/ Creating missing values in factors]
# OR
dplyr::select(TibbleObject, 2:5) #
</syntaxhighlight>


=== llply() ===
=== Concatenating two factor vectors ===
llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.
Not trivial. [https://stackoverflow.com/a/5068939 How to concatenate factors, without them being converted to integer level?].
<pre>
<pre>
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
</pre>
</pre>
where rLLID is a list of entrez ID. For example,
<pre>
get("org.Hs.egGO")[["6772"]]
</pre>
returns a list of 49 GOs.


=== ddply() ===
=== droplevels() ===
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/droplevels droplevels()]: drop unused levels from a factor or, more commonly, from factors in a data frame.


=== ldply() ===
=== factor(x , levels = ...) vs levels(x) <-  ===
[http://rpsychologist.com/an-r-script-to-automatically-look-at-pubmed-citation-counts-by-year-of-publication/ An R Script to Automatically download PubMed Citation Counts By Year of Publication]
<span style="color: red">Note [https://stat.ethz.ch/R-manual/R-devel/library/base/html/levels.html levels(x)] is to set/rename levels, not reorder.</span> Use <s>'''relevel()'''</s> or '''factor()''' to reorder.


== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
{| class="wikitable"
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
|-
| [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/levels levels()]</br>[https://www.rdocumentation.org/packages/plyr/versions/1.8.9/topics/revalue plyr::revalue()]</br>[https://rdocumentation.org/packages/forcats/versions/1.0.0/topics/fct_recode forcats::fct_recode()]
| rename levels
|-
| [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/factor factor(, levels)]
| reorder levels
|}


=== get_seed() ===
See the same blog
<syntaxhighlight lang='rsplus'>
<syntaxhighlight lang='rsplus'>
get_seed <- function() {
sizes <- factor(c("small", "large", "large", "small", "medium"))
  sample.int(.Machine$integer.max, 1)
sizes
}
#> [1] small  large  large  small  medium
#> Levels: large medium small
 
sizes2 <- factor(sizes, levels = c("small", "medium", "large")) # reorder levels but data is not changed
sizes2
# [1] small  large  large  small  medium
# Levels: small medium large
 
sizes3 <- sizes
levels(sizes3) <- c("small", "medium", "large") # rename, not reorder
                                                # large -> small
                                                # medium -> medium
                                                # small -> large
sizes3
# [1] large  small  small  large  medium
# Levels: small medium large
</syntaxhighlight>
A regression example.
<syntaxhighlight lang='rsplus'>
set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  0.96804    0.06610  14.65  <2e-16 ***
# x2          0.99620    0.09462  10.53  <2e-16 ***
 
# Wrong way when we want to change the baseline level to '2'
# No change on the model fitting except the apparent change on the variable name in the printout
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  0.96804    0.06610  14.65  <2e-16 ***
# x1          0.99620    0.09462  10.53  <2e-16 ***
 
# Correct way if we want to change the baseline level to '2'
# The estimate was changed by flipping the sign from the original data
sample_data$x <- relevel(x, ref = "2")
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  1.96425    0.06770  29.01  <2e-16 ***
# x1          -0.99620    0.09462  -10.53  <2e-16 ***
</syntaxhighlight>
</syntaxhighlight>


== How to select a seed for simulation or randomization ==
=== stats::relevel() ===
[https://sciprincess.wordpress.com/2019/03/14/how-to-select-a-seed-for-simulation-or-randomization/ How to select a seed for simulation or randomization]
[https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/relevel relevel]. This function can only be used to change the '''reference level''' of a factor variable. '''It does not directly create an arbitrary order of levels'''. That is, it is useful in lm() or aov(), etc.
 
=== reorder(), levels() and boxplot() ===
<ul>
<li>[https://www.r-bloggers.com/2023/09/how-to-reorder-boxplots-in-r-a-comprehensive-guide/ How to Reorder Boxplots in R: A Comprehensive Guide] (tapply() method, simple & effective)
<li>[https://stat.ethz.ch/R-manual/R-devel/library/stats/html/reorder.factor.html reorder()].This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
<pre>
# Syntax:
# newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order


== set.seed(), for loop and saving random seeds ==
(bymedian <- with(InsectSprays, reorder(spray, count, median)) )
http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
class(bymedian)
levels(bymedian)
boxplot(count ~ bymedian, data = InsectSprays,
        xlab = "Type of spray", ylab = "Insect count",
        main = "InsectSprays data", varwidth = TRUE,
        col = "lightgray") # boxplots are sorted according to the new levels
boxplot(count ~ spray, data = InsectSprays,
        xlab = "Type of spray", ylab = "Insect count",
        main = "InsectSprays data", varwidth = TRUE,
        col = "lightgray") # not sorted
</pre>
<li>[http://www.deeplytrivial.com/2020/05/statistics-sunday-my-2019-reading.html Statistics Sunday: My 2019 Reading] (reorder function)
</ul>


<syntaxhighlight lang='rsplus'>
=== factor() vs ordered() ===
set.seed(1001)  
<pre>
data <- vector("list", 30)
factor(levels=c("a", "b", "c"), ordered=TRUE)
seeds <- vector("list", 30)  
# ordered(0)
for(i in 1:30) {
# Levels: a < b < c
  seeds[[i]] <- .Random.seed
  data[[i]] <- runif(5)  
}
.Random.seed <- seeds[[23]]  # restore
data.23 <- runif(5)
data.23
data[[23]]
</syntaxhighlight>
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.


== sample() inaccurate on very large populations, fixed in R 3.6.0 ==
factor(levels=c("a", "b", "c"))
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17494 The default method for generating from a discrete uniform distribution (used in ‘sample()’, for instance) has been changed]. In prior versions, the probability of generating each integer could vary from equal by up to 0.04% (or possibly more if generating more than a million different integers). See also [https://www.r-bloggers.com/whats-new-in-r-3-6-0/amp/ What's new in R 3.6.0] by David Smith. <syntaxhighlight lang='rsplus'>
# factor(0)
# R 3.5.3
# Levels: a b c
set.seed(123)
 
m <- (2/5)*2^32
ordered(levels=c("a", "b", "c"))
m > 2^31
# Error in factor(x, ..., ordered = TRUE) :
# [1] FALSE
#  argument "x" is missing, with no default
log10(m)
</pre>
# [1] 9.23502
 
x <- sample(m, 1000000, replace = TRUE)
== Data frame ==
table(x %% 2)
* http://adv-r.had.co.nz/Data-structures.html#data-frames. '''A data frame is a list of equal-length vectors'''. So a data frame is not a vector nor a matrix though it looks like a matrix.
#      0      1
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/
# 400070 599930
</syntaxhighlight>
* [https://blog.daqana.com/en/fast-sampling-support-in-dqrng/ Fast sampling support in dqrng]
* Differences of the output of sample() <syntaxhighlight lang='rsplus'>
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5


# R 3.6.0
=== stringsAsFactors = FALSE ===
# docker run --net=host -it --rm r-base:3.6.0
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5


</syntaxhighlight>
We can use '''options(stringsAsFactors=FALSE)''' forces R to import character data as character objects.


== Regular Expression ==
In R 4.0.0, [https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/ stringAsFactors=FALSE] will be default. This also affects read.table() function.
See [[Regular_expression|here]].


== Read rrd file ==
=== check.names = FALSE ===
* https://en.wikipedia.org/wiki/RRDtool
Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.
* http://oss.oetiker.ch/rrdtool/
<pre>
* https://github.com/pldimitrov/Rrd
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/
  1a 2a
1  1  1
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1  1  1
2  2  2
</pre>


== file, connection ==
=== Create unique rownames: make.unique() ===
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
<pre>
* read() and write()
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
* read.table() and write.table()
rownames(mydf) <- make.unique(groupCodes)
<syntaxhighlight lang='rsplus'>
</pre>
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)
</syntaxhighlight>


== Clipboard (?connections), textConnection(), pipe() ==
=== data.frame() will change rownames ===
* On Windows, we can use readClipboard() and writeClipboard(). <syntaxhighlight lang='rsplus'>
<pre>
source("clipboard")
class(df2)
read.table("clipboard")
# [1] "matrix" "array"
</syntaxhighlight>
rownames(df2)[c(9109, 44999)]
* reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function: <syntaxhighlight lang='rsplus'>
# [1] "A1CF"     "A1BG-AS1"
x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
rownames(data.frame(df2))[c(9109, 44999)]
# Or on Mac
# [1] "A1CF"     "A1BG.AS1"
x <- read.delim(pipe("pbpaste"))
</pre>
# safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
</syntaxhighlight> An example is to copy data from [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others?answertab=active#tab-top this post]. In this case we need to use read.table() instead of read.delim().
* Write to clipboard on mac. Note: '''pbcopy''' and '''pbpaste''' are macOS terminal commands. See [http://osxdaily.com/2007/03/05/manipulating-the-clipboard-from-the-command-line/ pbcopy & pbpaste: Manipulating the Clipboard from the Command Line].
** pbcopy: takes standard input and places it in the clipboard buffer
** pbpaste: takes data from the clipboard buffer and writes it to the standard output
: <syntaxhighlight lang='rsplus'>
clip <- pipe("pbcopy", "w")
write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
close(clip)
</syntaxhighlight>
* On Linux, we need to install "xclip". See [https://stackoverflow.com/questions/45799496/r-copy-from-clipboard-in-ubuntu-linux R Copy from Clipboard in Ubuntu Linux]. It seems to work. <syntaxhighlight lang='rsplus'>
# sudo apt-get install xclip
read.table(pipe("xclip -selection clipboard -o",open="r"))
</syntaxhighlight>


== read/manipulate binary data ==
=== Print a data frame without rownames ===
* x <- readBin(fn, raw(), file.info(fn)$size)
<pre>
* rawToChar(x[1:16])
# Method 1.  
* See Biostrings C API
rownames(df1) <- NULL


== String Manipulation ==
# Method 2.  
* [http://gastonsanchez.com/blog/resources/how-to/2013/09/22/Handling-and-Processing-Strings-in-R.html ebook] by Gaston Sanchez.
print(df1, row.names = FALSE)
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
</pre>
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]


== HTTPs connection ==  
=== Convert data frame factor columns to characters ===
HTTPS connection becomes default in R 3.2.2. See
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
* http://blog.rstudio.org/2015/08/17/secure-https-connections-for-r/
{{Pre}}
* http://blog.revolutionanalytics.com/2015/08/good-advice-for-security-with-r.html
# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)


[http://developer.r-project.org/blosxom.cgi/R-devel/2016/12/15#n2016-12-15 R 3.3.2 patched] The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)
# Method 2:
bob[] <- lapply(bob, as.character)
</pre>


== setInternet2 ==
[https://stackoverflow.com/a/2853231 To replace only factor columns]:
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.
<pre>
# Method 1:
i <- sapply(bob, is.factor)
bob[i] <- lapply(bob[i], as.character)


Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
# Method 2:
<pre>
library(dplyr)
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
bob %>% mutate_if(is.factor, as.character) -> bob
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)
</pre>
</pre>
It seems the bug was fixed in R 3.2-branch. See [https://github.com/wch/r-source/commit/3a02ed3a50ba17d9a093b315bf5f31ffc0e21b89 8/16/2015] patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385098%28v=vs.85%29.aspx InternetOpenUrl()] function of [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385473%28v=vs.85%29.aspx wininet] library. [http://slacksite.com/other/ftp.html This article] and [http://stackoverflow.com/questions/1699145/what-is-the-difference-between-active-and-passive-ftp this post] explain differences of active and passive FTP.


The following R command will show the exact svn revision for the R you are currently using.
=== Sort Or Order A Data Frame ===
[https://howtoprogram.xyz/2018/01/07/r-how-to-order-a-data-frame/ How To Sort Or Order A Data Frame In R]
# df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
# library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
# library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
# library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)
 
=== data.frame to vector ===
<pre>
<pre>
R.Version()$"svn rev"
df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))
 
class(df)
# [1] "data.frame"
class(t(df))
# [1] "matrix" "array"
class(unlist(df))
# [1] "numeric"
 
# Method 1: Convert data frame to matrix using as.matrix()
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)  # [1] 1 2 3 4 5 6
vec2 <- c(mat)
 
# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))
 
# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6
 
# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3
#  1  2  3  4  5  6
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6
</pre>
</pre>
Q: Why as.vector(df) cannot convert a data frame into a vector?


If setInternet2(T), then https protocol is supported in download.file().  
A: The as.vector function cannot be used directly on a data frame to convert it into a vector because a data frame is a list of vectors (i.e., its columns) and '''as.vector only removes the attributes of an object to create a vector'''. When you apply as.vector to a data frame, R does not know how to concatenate these independent columns (which could be of different types) into a single vector. Therefore, it doesn’t perform the operation. Therefore as.vector() returns the underlying list structure of the data frame instead of converting it into a vector.
 
However, when you transpose the data frame using t(), it gets converted into a matrix. A matrix in R is a vector with dimensions. Therefore, all elements of the matrix must be of the same type. If they are not, R will coerce them to be so. Once you have a matrix, as.vector() can easily convert it into a vector because all elements are of the same type.
 
=== Using cbind() to merge vectors together? ===
It’s a common mistake to try and create a data frame by cbind()ing vectors together. This doesn’t work because cbind() will create a matrix unless one of the arguments is already a data frame. Instead use data.frame() directly. See [http://adv-r.had.co.nz/Data-structures.html#data-frames Advanced R -> Data structures] chapter.  


When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.  
=== cbind NULL and data.frame ===
[https://9to5tutorial.com/cbind-can-t-combine-null-with-dataframe cbind can't combine NULL with dataframe]. Add as.matrix() will fix the problem.


The setInternet2() function is defined in [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].
=== merge ===
* [https://thomasadventure.blog/posts/r-merging-datasets/ All You Need To Know About Merging (Joining) Datasets in R]. If we like to merge/join by the rownames, we can use '''dplyr::rownames_to_column()'''; see [https://stackoverflow.com/a/42418771 dplyr left_join() by rownames].
* [https://www.geeksforgeeks.org/merge-dataframes-by-row-names-in-r/ Merge DataFrames by Row Names in R]
* [https://jozefhajnala.gitlab.io/r/r006-merge/ How to perform merges (joins) on two or more data frames with base R, tidyverse and data.table]
* [https://www.dummies.com/programming/r/how-to-use-the-merge-function-with-data-sets-in-r/ How to understand the different types of merge]


'''R up to 3.2.2'''
Special character in the matched variable can create a trouble when we use merge() or dplyr::inner_join(). I guess R internally turns df2 (a matrix but not a data frame) to a data frame (so rownames are changed if they contain special character like "-"). This still does not explain the situation when I
<pre>
<pre>
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
class(df1); class(df2)
# [1] "data.frame"  # 2 x 2
# [1] "matrix" "array" # 52439 x 2
rownames(df1)
# [1] "A1CF"    "A1BG-AS1"
merge(df1, df2[c(9109, 44999), ], by=0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BG-AS1    0    0  7.321358  6.908333
# 2      A1CF    0    0  3.011470  1.189578
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1      A1CF    0    0    3.01147  1.189578
rownames(df2)[38958:38959]
# [1] "ITFG2-AS1"  "ADGRD1-AS1"
 
rownames(df1)[2] <- "A1BGAS1"
rownames(df2)[44999] <- "A1BGAS1"
merge(df1, df2, by= 0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BGAS1    0    0  7.321358  6.908333
# 2      A1CF    0    0  3.011470  1.189578
</pre>
</pre>
See also
* <src/include/Internal.h> (declare do_setInternet2()),
* <src/main/names.c> (show do_setInternet2() in C)
* <src/main/internet.c>  (define do_setInternet2() in C).


Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file. If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2,  it is also possible to download from https by setIneternet2(T).
=== is.matrix: data.frame is not necessarily a matrix ===
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/matrix ?matrix]. is.matrix returns TRUE '''if x is a vector and has a "dim" attribute of length 2''' and FALSE otherwise.  


'''R 3.3.0'''
An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by
<pre>
<pre>
setInternet2 <- function(use = TRUE) {
X <- data.frame(x=1:2, y=3:4)
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>
</pre>
The 'X' object is NOT a vector and it does NOT have the "dim" attribute. It has only 3 attributes: "names", "row.names" & "class". Note that dim() function works fine and returns correctly though there is not "dim" attribute.


Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.
Another example that is a data frame but not a matrix is the built-in object ''cars''; see ?matrix. It is not a vector
 
=== Convert a data frame to a matrix: as.matrix() vs data.matrix() ===
If I have a data frame X which recorded the time of some files.


== File operation ==
* is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
* list.files()
* as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
* file.info()
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/data.matrix data.matrix(X)] will convert the time to numerical values. So use data.matrix() if the data is numeric. The returned object is not a data frame anymore.
* dir.create()
* file.create()
* file.copy()


== read/download/source a file from internet ==
<syntaxhighlight lang='r'>
=== Simple text file http ===
# latex directory contains cache files from knitting an rmarkdown file
<pre>
X <- list.files("latex/", full.names = T) %>%
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
    grep("RData", ., value=T) %>%
</pre>
    file.info() %>% 
    `[`("mtime")
X %>% is.data.frame() # TRUE
X %>% is.matrix() # FALSE
X %>% as.matrix() %>% is.matrix() # TRUE
X %>% data.matrix() %>% is.matrix() # TRUE
X %>% as.matrix() %>% "["(1:2, ) # timestamps
X %>% data.matrix() %>% "["(1:2, ) # numeric
</syntaxhighlight>


=== Zip file and url() function ===
* The '''as.matrix()''' function is used to coerce an object into a matrix. It can be used with various types of R objects, such as vectors, data frames, and arrays.
<pre>
* The '''data.matrix()''' function is specifically designed for converting a data frame into a matrix by coercing all columns to numeric values. If the data frame contains non-numeric columns, such as character or factor columns, data.matrix() will convert them to numeric values if possible (e.g., by converting factors to their integer codes).
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
* See the following example where as.matrix() and data.matrix() return different resuls.
source(con)
<syntaxhighlight lang='r'>
close(con)
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
</pre>
mat <- as.matrix(df)
Here url() function is like file(), gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.
mat
#      a  b 
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array"
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array"
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"
</syntaxhighlight>


Another example of using url() is
=== matrix vs data.frame ===
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
<pre>
<pre>
load(url("http:/www.example.com/example.RData"))
Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
[1] "time"  "status" "treat"  "AKT1"  "BRAF"  "FLOT2"  "MTOR"  "PCK2"  "PIK3CA"
[10] "RAF1" 
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10
</pre>
</pre>


=== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ===
Case 2:
This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.
{{Pre}}
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])   # OK


=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
<pre>
unique(ip2$Priority)     # OK
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))
</pre>
</pre>


=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
The length of a matrix and a data frame is different.
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]
{{Pre}}
> length(matrix(1:6, 3, 2))
[1] 6
> length(data.frame(matrix(1:6, 3, 2)))
[1] 2
> x[1]
  X1
1  1
2  2
3  3
4  4
5  5
6  6
> x[[1]]
[1] 1 2 3 4 5 6
</pre>
So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.
 
=== How to Remove Duplicates ===
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]
 
=== Convert a matrix (not data frame) of characters to numeric ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
{{Pre}}
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
    a    b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)    # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"


=== Github files https using RCurl package ===
> mode(tmp) <- "numeric"
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
> sum(tmp)
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
[1] 1.917
<pre>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
</pre>
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package


== summarytools: create summary tables for vectors and data frames ==
=== Convert Data Frame Row to Vector ===
https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.
as.numeric() or '''c()'''
 
=== Convert characters to integers ===
mode(x) <- "integer"


== Create publication tables using '''tables''' package ==
=== Non-Standard Evaluation ===
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]
[https://thomasadventure.blog/posts/understanding-nse-part1/ Understanding Non-Standard Evaluation. Part 1: The Basics]


R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,
=== Select Data Frame Columns in R ===
<syntaxhighlight lang='rsplus'>
This is part of series of DATA MANIPULATION IN R from [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ datanovia.com]
> library(tables)
 
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
* select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
                                                 
* select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
                Sepal.Length      Sepal.Width   
* Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names
Species    n  mean        sd  mean        sd 
 
  setosa      50 5.01        0.35 3.43        0.38
Another way is to the dollar sign '''$''' operator (?"$") to extract rows or column from a data frame.
versicolor  50 5.94         0.52 2.77        0.31
<pre>
virginica   50 6.59        0.64 2.97        0.32
class(USArrests) # "data.frame"
All        150 5.84        0.83 3.06        0.44
USArrests$"Assault"
> str(iris)
</pre>
'data.frame':  150 obs. of  5 variables:
Note that for both data frame and matrix objects, we need to use the '''[''' operator to extract columns and/or rows.
$ Sepal.Length: num 5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
<pre>
$ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
$ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
#         Murder Assault
$ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
# Alabama   13.2     236
$ Species    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
# Alaska    10.0     263
USArrests[c("Murder", "Assault")]  # all rows
 
tmp <- data(package="datasets")
class(tmp$results) # "matrix" "array"
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix
</pre>
Note for a '''data.table''' object, we can extract columns using the column names without double quotes.
<pre>
data.table(USArrests)[1:2, list(Murder, Assault)]
</pre>
</pre>
and
 
=== Add columns to a data frame ===
[https://datasciencetut.com/how-to-add-columns-to-a-data-frame-in-r/ How to add columns to a data frame in R]
 
=== Exclude/drop/remove data frame columns ===
* [https://datasciencetut.com/remove-columns-from-a-data-frame/ How to Remove Columns from a data frame in R]
* [https://www.listendata.com/2015/06/r-keep-drop-columns-from-data-frame.html R: keep / drop columns from data frame]
<pre>
<pre>
# This example shows some of the less common options       
# method 1
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
df = subset(mydata, select = -c(x,z) )
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
 
> z <- rnorm(100)+5
# method 2
> fmt <- function(x) {
drop <- c("x","z")
  s <- format(x, digits=2)
df = mydata[,!(names(mydata) %in% drop)]
  even <- ((1:length(s)) %% 2) == 0
 
  s[even] <- sprintf("(%s)", s[even])
# method 3: dplyr
   s
mydata2 = select(mydata, -a, -x, -y)
}
mydata2 = select(mydata, -c(a, x, y))
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
mydata2 = select(mydata, -a:-y)
                  Status             
mydata2 = mydata[,!grepl("^INC",names(mydata))]
Sex    Statistic high  low    medium
</pre>
Female mean      4.88  4.96  5.17
 
        sd        (1.20) (0.82) (1.35)
=== Remove Rows from the data frame ===
  Male  mean      4.45  4.31  5.05
[https://datasciencetut.com/remove-rows-from-the-data-frame-in-r/ Remove Rows from the data frame in R]
        sd        (1.01) (0.93) (0.75)
 
</syntaxhighlight>
=== Danger of selecting rows from a data frame ===
<pre>
> dim(cars)
[1] 50  2
> data.frame(a=cars[1,], b=cars[2, ])
   a.speed a.dist b.speed b.dist
1      4      2      4    10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a b
speed 4 4
dist  2 10
</pre>


== ClinReport: Statistical Reporting in Clinical Trials ==
=== Creating data frame using structure() function ===
https://cran.r-project.org/web/packages/ClinReport/index.html
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]


== Append figures to PDF files ==
=== Create an empty data.frame ===
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
https://stackoverflow.com/questions/10689055/create-an-empty-data-frame
<pre>
# the column types default as logical per vector(), but are then overridden
a = data.frame(matrix(vector(), 5, 3,
              dimnames=list(c(), c("Date", "File", "User"))),
              stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)


== Extracting tables from PDFs ==
# similar to above
* [http://datascienceplus.com/extracting-tables-from-pdfs-in-r-using-the-tabulizer-package/ extracting Tables from PDFs in R] using Tabulizer. This needs the [https://cran.r-project.org/web/packages/rJava/index.html rJava] package. Linux works fine. Some issue came out on my macOS 10.12 Sierra. '''Library not loaded: /Library/Java/JavaVirtualMachines/jdk-9.jdk/Contents/Home/lib/server/libjvm.dylib. Referenced from: /Users/XXXXXXX/Library/R/3.5/library/rJava/libs/rJava.so'''.
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))
* [https://cran.r-project.org/web/packages/pdftools/ pdftools] - Text Extraction, Rendering and Converting of PDF Documents. [https://ropensci.org/technotes/2018/12/14/pdftools-20/ pdf_text() and pdf_data()] functions. <syntaxhighlight lang='rsplus'>
library(pdftools)
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
txt <- pdf_text(pdf_file) # length = number of pages
# Suppose the table we are interested in is on page 1
cat(txt[1]) # Good but not in a data frame format


pdf_data(pdf_file)[[1]]  # data frame/tibble format
# different data type
</syntaxhighlight> However, it seems it does not work on [http://www.bloodjournal.org/content/109/8/3177/tab-figures-only Table S6]. Tabulizer package is better at this case.
a <- data.frame(x1 = character(),
* [https://www.linuxuprising.com/2019/05/how-to-convert-pdf-to-text-on-linux-gui.html?m=1 How To Convert PDF To Text On Linux (GUI And Command Line)]. It works when I tested my PDF file. <syntaxhighlight lang='bash'>
                x2 = numeric(),
sudo apt install poppler-utils
                x3 = factor(),
pdftotext -layout input.pdf output.txt
                stringsAsFactors = FALSE)
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</pre>
</syntaxhighlight>


== Create flat tables in R console using ftable() ==
=== Objects from subsetting a row in a data frame vs matrix ===
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
> ftable(Titanic, row.vars = 1:3)
<ul>
                  Survived  No Yes
<li>Subsetting creates repeated rows. This will create unexpected rownames.
Class Sex    Age                 
<pre>
1st   Male  Child            0  5
R> z <- data.frame(x=1:3, y=2:4)
            Adult          118  57
R> rownames(z) <- letters[1:3]
      Female Child            0   1
R> rownames(z)[c(1,1)]
            Adult            4 140
[1] "a" "a"
2nd  Male  Child            0  11
R> rownames(z[c(1,1),])
            Adult          154  14
[1] "a"   "a.1"
      Female Child            0  13
R> z[c(1,1), ]
            Adult          13  80
    x y
3rd  Male  Child          35 13
a   1 2
            Adult          387 75
a.1 1 2
      Female Child          17 14
</pre>
            Adult          89 76
</li>
Crew Male  Child            0   0
<li>[https://stackoverflow.com/a/2545548 Convert a dataframe to a vector (by rows)] The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
            Adult          670 192
{{Pre}}
      Female Child            0   0
str(trainData)
            Adult            3 20
# 'data.frame': 503 obs. of 500 variables:
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
# $ bm001: num 0.429 1 -0.5 1.415 -1.899 ...
            Survived  No Yes
# $ bm002: num 0.0568 1 0.5 0.3556 -1.16 ...
Class Sex                   
# ...
1st   Male            118  62
trainData[1:3, 1:3]
      Female            4 141
#        bm001      bm002    bm003
2nd   Male            154  25
# 1  0.4289449 0.05676296 1.657966
      Female          13  93
# 2  1.0000000 1.00000000 1.000000
3rd  Male            422  88
# 3 -0.5000000 0.50000000 0.500000
      Female          106  90
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
Crew  Male            670 192
# Warning message:
       Female            3 20
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
#  row names were found from a short variable and have been discarded
            Survived  No Yes
</pre>
Sex    Class               
 
Male   1st            118  62
'trees' data from the 'datasets' package
       2nd            154  25
<pre>
       3rd           422  88
trees[1:3,]
      Crew          670 192
#  Girth Height Volume
Female 1st              4 141
# 1  8.3    70   10.3
       2nd            13  93
# 2  8.6    65  10.3
       3rd           106  90
# 3  8.8    63   10.2
      Crew            3 20
 
> str(Titanic)
# Wrong ways:
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
data.frame(trees[1,] , trees[2,])
- attr(*, "dimnames")=List of 4
#  Girth Height Volume Girth.1 Height.1 Volume.1
   ..$ Class   : chr [1:4] "1st" "2nd" "3rd" "Crew"
# 1  8.3    70  10.3    8.6       65    10.3
   ..$ Sex    : chr [1:2] "Male" "Female"
data.frame(time=trees[1,] , status=trees[2,])
   ..$ Age    : chr [1:2] "Child" "Adult"
#   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
   ..$ Survived: chr [1:2] "No" "Yes"
# 1       8.3          70       10.3          8.6           65          10.3
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
data.frame(time=as.vector(trees[1,]) , status=as.vector(trees[2,]))
> x
#  time.Girth time.Height time.Volume status.Girth status.Height status.Volume
          gear  3 4  5
# 1       8.3          70       10.3          8.6           65          10.3
cyl vs am             
data.frame(time=c(trees[1,]) , status=c(trees[2,]))
4   0  0        0  0  0
# time.Girth time.Height time.Volume status.Girth status.Height status.Volume
      1       0  0  1
# 1       8.3          70        10.3          8.6            65          10.3
    1 0        1  2  0
 
      1       0  6 1
# Right ways:
6   0  0        0  0  0
# method 1: dropping row names
      1        0  2  1
data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,])))
    1 0        2  2  0
# OR
      1        0 0  0
data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
8  0 0      12  0  0
#  time status
      1       0  0  2
# 1  8.3    8.6
    1 0        0  0  0
# 2 70.0   65.0
      1        0  0  0
# 3 10.3   10.3
> ftable(x, row.vars = c(2, 4))
# method 2: keeping row names
        cyl  4    6    8 
data.frame(time=t(trees[1,]) , status=t(trees[2,]))
        am  0  1 1 1
#          X1  X2
vs gear                     
# Girth   8.3  8.6
3         0  0  0  0 12  0
# Height 70.0 65.0
  4        0  0  0  2 0  0
# Volume 10.3 10.3
   5        0  1 1 0  2
data.frame(time=unlist(trees[1,]) , status=unlist(trees[2,]))
1 3         1 2 0  0  0
#        time status
   4         2 2 0  0  0
# Girth   8.3    8.6
   5         0  1  0  0  0  0
# Height 70.0   65.0
>
# Volume 10.3  10.3
> ## Start with expressions, use table()'s "dnn" to change labels
 
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
# Method 3: convert a data frame to a matrix
        dnn = c("Cylinders", "V/S", "Transmission", "Gears"))
is.matrix(trees)
# [1] FALSE
trees2 <- as.matrix(trees)
data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
#        time status
# Girth  8.3   8.6
# Height 70.0  65.0
# Volume 10.3   10.3
 
dim(trees[1,])
# [1] 1 3
dim(trees2[1, ])
# NULL
trees[1, ] # notice the row name '1' on the left hand side
#   Girth Height Volume
# 1   8.3    70  10.3
trees2[1, ]
# Girth Height Volume
#    8.3   70.0   10.3
</pre>
</li>
</ul>
 
=== Convert a list to data frame ===
[https://www.statology.org/convert-list-to-data-frame-r/ How to Convert a List to a Data Frame in R].
<pre>
# method 1
data.frame(t(sapply(my_list,c)))
 
# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)
 
# method 3
library(data.table)
rbindlist(my_list)
</pre>
 
=== tibble and data.table ===
* [[R#tibble | tibble]]
* [[Tidyverse#data.table|data.table]]
 
=== Clean a dataset ===
[https://finnstats.com/index.php/2021/04/04/how-to-clean-the-datasets-in-r/ How to clean the datasets in R]
 
== matrix ==
 
=== Define and subset a matrix ===
* [https://www.tutorialkart.com/r-tutorial/r-matrix/ Matrix in R]
** It is clear when a vector becomes a matrix the data is transformed column-wisely ('''byrow''' = FALSE, by default).
** When subsetting a matrix, it follows the format: '''X[rows, colums]''' or '''X[y-axis, x-axis]'''.  
 
<pre>
data <- c(2, 4, 7, 5, 10, 1)
A <- matrix(data, ncol = 3)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7  10
# [2,]    4    5    1
 
A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]   7  10
</pre>
 
=== Prevent automatic conversion of single column to vector ===
use '''drop = FALSE''' such as mat[, 1, drop = FALSE].
 
=== complete.cases(): remove rows with missing in any column ===
It works on a sequence of vectors, matrices and data frames.
 
=== NROW vs nrow ===
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/nrow ?nrow]. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.
 
=== matrix (column-major order) multiply a vector ===
* Matrices in R [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) are stored in a column-major order]. It means array slice A[,1] are contiguous.
 
{{Pre}}
> matrix(1:6, 3,2)
    [,1] [,2]
[1,]    1   4
[2,]    2    5
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3) # c(1,2,3) will be recycled to form a matrix. Good quiz.
    [,1] [,2]
[1,]    1    4
[2,]    4  10
[3,]    9  18
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
    [,1] [,2]
[1,]    1  16
[2,]    4    5
[3,]    9  12
</pre>
 
* [https://stackoverflow.com/a/20596490 How to divide each row of a matrix by elements of a vector in R]
 
=== add a vector to all rows of a matrix ===
[https://stackoverflow.com/a/39443126 add a vector to all rows of a matrix]. sweep() or rep() is the best.


          Cylinders    4    6    8 
=== sparse matrix ===
          Transmission  0  1  0  1  0  1
[https://stackoverflow.com/a/10555270 R convert matrix or data frame to sparseMatrix]
V/S Gears                             
 
0  3                  0  0  0  0 12  0
To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, '''as.vector()'''.
    4                  0  0  0  2  0  0
 
    5                  0  1  0  1  0  2
== Attributes ==
1  3                  1  0  2  0  0  0
* [https://statisticaloddsandends.wordpress.com/2020/10/19/attributes-in-r/ Attributes in R]
    4                  2  6  2  0  0  0
* [http://adv-r.had.co.nz/Data-structures.html#attributes Data structures] in "Advanced R"
    5                  0  1  0  0  0  0
 
</syntaxhighlight>
== Names ==
[https://masalmon.eu/2023/11/06/functions-dealing-with-names/ Useful functions for dealing with object names]. (Un)Setting object names: stats::setNames(), unname() and rlang::set_names()


=== [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins] ===
=== Print a vector by suppressing names ===
Puts Arbitrary Margins On Multidimensional Tables Or Arrays
Use '''unname'''. sapply(, , USE.NAMES = FALSE).


== tracemem, data type, copy ==
== format.pval/print p-values/format p values ==
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]
[https://rdrr.io/r/base/format.pval.html format.pval()]. By default it will show 5 significant digits (getOption("digits")-2).
{{Pre}}
> format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"   
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"


== Tell if the current R is running in 32-bit or 64-bit mode ==
R> pvalue
<pre>
[1] 0.0004632104
8 * .Machine$sizeof.pointer
R> print(pvalue, digits =20)
[1] 0.00046321036188223807528
R> format.pval(pvalue)
[1] "0.00046321"
R> format.pval(pvalue * 1e-1)
[1] "4.6321e-05"
R> format.pval(0.00004632)
[1] "4.632e-05"
R> getOption("digits")
[1] 7
</pre>
</pre>
where '''sizeof.pointer''' returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.


== 32- and 64-bit ==
== Customize R: options() ==
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
* Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).


== Handling length 2^31 and more in R 3.0.0 ==
=== Change the default R repository, my .Rprofile ===
[[Rstudio#Change_repository|Change R repository]]


From R News for 3.0.0 release:
Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local '''.Rprofile''' settings take precedence.
 
''There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.
''


In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error
For example, I can specify the R mirror I like by creating a single line '''.Rprofile''' file under my home directory. Another good choice of repository is '''cloud.r-project.org'''.
<pre>
> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector
</pre>


However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
Type '''file.edit("~/.Rprofile")'''
{{Pre}}
local({
  r = getOption("repos")
  r["CRAN"] = "https://cran.rstudio.com/"
  options(repos = r)
})
options(continue = "  ", editor = "nano")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
  .Last <- function() try(savehistory("~/.Rhistory"))
}
</pre>
 
=== Change the default web browser for utils::browseURL() ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put
<pre>
<pre>
> system.time(x <- seq(1,2^31))
options(browser='seamonkey')
  user  system elapsed
  8.604  11.060 120.815
> length(x)
[1] 2147483648
> length(x)/2^20
[1] 2048
> gc()
            used    (Mb) gc trigger    (Mb)  max used    (Mb)
Ncells    183823    9.9    407500    21.8    350000    18.7
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
>
</pre>
</pre>
Note:
in the '''.Rprofile''' of your home directory. If the browser is not in the global PATH, we need to put the full path above.
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].


== NA in index ==
For one-time only purpose, we can use the ''browser'' option in help.start() function:
* Question: what is seq(1, 3)[c(1, 2, NA)]?
{{Pre}}
> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
</pre>


Answer: It will reserve the element with NA in indexing and return the value NA for it.
We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
* [https://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html


* Question: What is TRUE & NA?
=== Change the default editor ===
Answer: NA
On my Linux and mac, the default editor is "vi". To change it to "nano",
{{Pre}}
options(editor = "nano")
</pre>


* Question: What is FALSE & NA?
=== Change prompt and remove '+' sign ===
Answer: FALSE
See https://stackoverflow.com/a/1448823.
{{Pre}}
options(prompt="R> ", continue=" ")
</pre>


* Question: c("A", "B", NA) != "" ?
=== digits ===
Answer: TRUE TRUE NA
* [https://gist.github.com/arraytools/26a0b359541f4fc9fddc8f0a0c94489e Read and compute the sum of a numeric matrix file] using R vs Python vs C++. Note by default R does not show digits after the decimal point because the number is large.
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
* [https://stackoverflow.com/a/10712012 ?print.default]
* [https://stackoverflow.com/a/12135122 Formatting Decimal places in R, round()]. [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/format format()] where '''nsmall''' controls the minimum number of digits to the right of the decimal point
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17668 numerical error in round() causing round to even to fail] 2019-12-05
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Round signif()] rounds x to n significant digits.
<pre>
R> signif(pi, 3)
[1] 3.14
R> signif(pi, 5)
[1] 3.1416
</pre>
</li>
</ul>
* The default digits 7 may be too small. For example, '''if a number is very large, then we may not be able to see (enough) value after the decimal point'''. The acceptable range is 1-22. See the following examples


* Question: which(c("A", "B", NA) != "") ?
In R,
Answer: 1 2
{{Pre}}
> options()$digits # Default
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
[1] 100000.1
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11
</pre>


* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
In Python,
Answer: TRUE TRUE FALSE
{{Pre}}
>>> 100000.07 + .04
100000.11
</pre>


* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
Answer: TRUE TRUE FALSE
[https://datasciencetut.com/how-to-turn-off-scientific-notation-in-r/ How to Turn Off Scientific Notation in R?]


'''Conclusion''': In order to exclude empty or NA for numerical or character data type, we can use '''which()''' or a convenience function '''keep.complete(x) <- function(x) x != "" & !is.na(x)'''. This will guarantee return logical values and not contain NAs.
This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.
{{Pre}}
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)  
[1] 3.164561e-01 2.970700e+04 9.387400e+04


Don't just use x != "" OR !is.na(x).
# Method 1. Without changing the global option
> format(c(numer/denom, numer, denom), scientific=FALSE)
[1] "    0.3164561" "29707.0000000" "93874.0000000"


== Constant and 'L' ==
# Method 2. Change the global option
Add 'L' after a constant. For example,
> options(scipen=999)
<syntaxhighlight lang='rsplus'>
> numer/denom
for(i in 1L:n) { }
[1] 0.3164561
 
> c(numer/denom, numer, denom)
if (max.lines > 0L) { }
[1]    0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]    0.8 29707.0 93874.0
</pre>
 
=== Suppress warnings: options() and capture.output() ===
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.
{{Pre}}
op <- options("warn")
options(warn = -1)
....
options(op)


label <- paste0(n-i+1L, ": ")
# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)
</pre>


n <- length(x);  if(n == 0L) { }
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/warning suppressWarnings()]
</syntaxhighlight>
<pre>
suppressWarnings( foo() )


== Factor ==
foo <- capture.output(
=== labels argument ===
bar <- suppressWarnings(
We can specify the factor levels and new labels using the factor() function.
{print( "hello, world" );
  warning("unwanted" )} ) )
</pre>


<syntaxhighlight lang='rsplus'>
[https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/capture.output capture.output()]
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
<pre>
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))
</pre>
</syntaxhighlight>


=== Create a factor from a continuous variable: cut() ===
=== Converts warnings into errors ===
<syntaxhighlight lang='rsplus'>
options(warn=2)
facVar <- cut(contVar, c(-3, 0, 2, 4), labels = c("low", "medium", "high"))
</syntaxhighlight>
Or [https://youtu.be/7oyiPBjLAWY?t=2480 a tibble object]
<syntaxhighlight lang='rsplus'>
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
      age_cat = case_when(
          age_yrs < 2 ~ "baby",
          age_yrs < 13 ~ "kid",
          age_yrs < 20 ~ "teen",
          TRUE        ~ "adult")
)
</syntaxhighlight>


== Data frame ==
=== demo() function ===
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/
<ul>
<li>[https://stackoverflow.com/a/18746519 How to wait for a keypress in R?] PS [https://stat.ethz.ch/R-manual/R-devel/library/base/html/readline.html readline()] is different from readLines().
<pre>
for(i in 1:2) { print(i); readline("Press [enter] to continue")}
</pre>
<li>Hit 'ESC' or Ctrl+c to skip the prompt "Hit <Return> to see next plot:" </li>
<li>demo() uses [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/options options()] to ask users to hit Enter on each plot
<pre>
op <- options(device.ask.default = ask)  # ask = TRUE
on.exit(options(op), add = TRUE)
</pre>
</li>
</ul>


=== stringsAsFactors = FALSE ===
== sprintf ==
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
=== paste, paste0, sprintf ===
[https://www.r-bloggers.com/paste-paste0-and-sprintf/ this post], [https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ 3 R functions that I enjoy]


=== Convert data frame factor columns to characters ===
=== sep vs collapse in paste() ===
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
* sep is used if we supply multiple input objects to paste()
<syntaxhighlight lang='rsplus'>
* collapse is used to make the output of length 1. It is commonly used if we have only 1 input object
# Method 1:
<pre>
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
R> paste("a", "A", sep=",")
[1] "a,A"
R> paste("a", "A", sep=",", collapse="-")
[1] "a,A"
R> paste(c("a", "A"), collapse="-")
[1] "a-A"


# Method 2:
R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
bob[] <- lapply(bob, as.character)
[1] "a,A - b,B - c,C"
</syntaxhighlight>
R> paste(letters[1:3], collapse = "-")
[1] "a-b-c"
</pre>


=== data.frame to vector ===
=== Format number as fixed width, with leading zeros ===
<syntaxhighlight lang='rsplus'>
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
> a= matrix(1:6, 2,3)
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1
> rownames(a) <- c("a", "b")
> colnames(a) <- c("x", "y", "z")
> a
  x y z
a 1 3 5
b 2 4 6
> unlist(data.frame(a))
x1 x2 y1 y2 z1 z2
1 2  3  4  5  6
</syntaxhighlight>


=== merge ===
{{Pre}}
[https://jozefhajnala.gitlab.io/r/r006-merge/ How to perform merges (joins) on two or more data frames with base R, tidyverse and data.table]
# sprintf()
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


=== matrix vs data.frame ===
# formatC()
<syntaxhighlight lang='rsplus'>
paste("name", formatC(a, width=3, flag="0"), sep="_")
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])  # OK


ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
# gsub()
unique(ip2$Priority)    # OK
paste0("bm", gsub(" ", "0", format(5:15)))
</syntaxhighlight>
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
</pre>


The length of a matrix and a data frame is different.
=== formatC and prettyNum (prettifying numbers) ===
<syntaxhighlight lang='rsplus'>
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/formatC formatC() & prettyNum()]
> length(matrix(1:6, 3, 2))
* [[R#format.pval|format.pval()]]
[1] 6
<pre>
> length(data.frame(matrix(1:6, 3, 2)))
R> (x <- 1.2345 * 10 ^ (-8:4))
[1] 2
[1] 1.2345e-08 1.2345e-07 1.2345e-06 1.2345e-05 1.2345e-04 1.2345e-03
> x[1]
[7] 1.2345e-02 1.2345e-01 1.2345e+00 1.2345e+01 1.2345e+02 1.2345e+03
   X1
[13] 1.2345e+04
1 1
R> formatC(x)
2  2
[1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "0.0001234" "0.001234"
3 3
[7] "0.01235"   "0.1235"    "1.234"    "12.34"    "123.4"    "1234"
4 4
[13] "1.234e+04"
5 5
R> formatC(x, digits=3)
6  6
  [1] "1.23e-08" "1.23e-07" "1.23e-06" "1.23e-05" "0.000123" "0.00123"
> x[[1]]
  [7] "0.0123"  "0.123"    "1.23"    "12.3"    " 123"    "1.23e+03"
[1] 1 2 3 4 5 6
[13] "1.23e+04"
</syntaxhighlight>
R> formatC(x, digits=3, format="e")
So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.
[1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "1.234e-04" "1.234e-03"
[7] "1.235e-02" "1.235e-01" "1.234e+00" "1.234e+01" "1.234e+02" "1.234e+03"
[13] "1.234e+04"
 
R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"
</pre>


=== Convert a matrix (not data frame) of characters to numeric ===
=== Format(x, scientific = TRUE) ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
Print numeric data in exponential format, so .0001 prints as 1e-4
<syntaxhighlight lang='rsplus'>
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
    a    b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)    # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"


> mode(tmp) <- "numeric"
== Creating publication quality graphs in R ==
> sum(tmp)
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
[1] 1.917
</syntaxhighlight>


=== Select Data Frame Columns in R ===
== HDF5 : Hierarchical Data Format==
This is part of series of DATA MANIPULATION IN R from [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ datanovia.com]
HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.


* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
* select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
* select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
* Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names
* rhdf5 is used by [http://amp.pharm.mssm.edu/archs4/data.html ARCHS4] where you can download R program that will download hdf5 file storing expression and metadata such as gene ID, sample/GSM ID, tissues, et al.
 
== Formats for writing/saving and sharing data ==
[http://www.econometricsbysimulation.com/2016/12/efficiently-saving-and-sharing-data-in-r_46.html Efficiently Saving and Sharing Data in R]


=== Creating data frame using structure() function ===
== Write unix format files on Windows and vice versa ==
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html


=== Warning: row names were found from a short variable and have been discarded ===
== with() and within() functions ==
https://stackoverflow.com/a/23534617
* [https://www.r-bloggers.com/2023/07/simplify-your-code-with-rs-powerful-functions-with-and-within/ Simplify Your Code with R’s Powerful Functions: with() and within()]
* within() is similar to with() except it is used to create new columns and merge them with the original data sets. But if we just want to create a new column, we can just use df$newVar = . The following example is from [http://www.youtube.com/watch?v=pZ6Bnxg9E8w&list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP youtube video].
<pre>
closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)


My example:
mk <- within(mariokart, {
<syntaxhighlight lang='rsplus'>
            closePr <- totalPr - shipPr
str(trainData)
     })
# 'data.frame': 503 obs. of  500 variables:
head(mk) # new column closePr
#  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
#  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
# ...
trainData[1:3, 1:3]
#        bm001     bm002    bm003
# 1  0.4289449 0.05676296 1.657966
# 2  1.0000000 1.00000000 1.000000
# 3 -0.5000000 0.50000000 0.500000
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
# Warning message:
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
#   row names were found from a short variable and have been discarded
</syntaxhighlight>


== matrix (column-major order) multiply a vector ==
mk <- mariokart
* [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) is following the column-major order]
aggregate(. ~ wheels + cond, mk, mean)
# create mean according to each level of (wheels, cond)


<syntaxhighlight lang='rsplus'>
aggregate(totalPr ~ wheels + cond, mk, mean)
> matrix(1:6, 3,2)
    [,1] [,2]
[1,]    1    4
[2,]    2    5
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3)
    [,1] [,2]
[1,]    1    4
[2,]    4  10
[3,]    9  18
> matrix(1:6, 3,2) * c(1,2,3,4)
    [,1] [,2]
[1,]    1  16
[2,]    4    5
[3,]    9  12
</syntaxhighlight>


== Print a vector by suppressing names ==
tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)
Use '''unname'''.
</pre>


== format.pval ==
== stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals ==
<syntaxhighlight lang='rsplus'>
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
> args(format.pval)
* [https://www.dataanalytics.org.uk/tally-plots-in-r/ Tally plots in R]
function (pv, digits = max(1L, getOption("digits") - 2L), eps = .Machine$double.eps,
* https://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
    na.form = "NA", ...)
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package


> format.pval(c(stats::runif(5), pi^-100, NA))
== Plot histograms as lines ==
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"   
https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.
> format.pval(c(0.1, 0.0001, 1e-27))
<pre>
[1] "1e-01"  "1e-04"  "<2e-16"
x2=invisible(hist(out2$EB))
</syntaxhighlight>
y2=invisible(hist(out2$Bench))
z2=invisible(hist(out2$EB0.001))


== options(digits) ==
plot(x=x2$mids, y=x2$density, type="l")
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
lines(y2$mids, y2$density, lty=2, pwd=2)
* [https://stackoverflow.com/a/10712012 ?print.default]
lines(z2$mids, z2$density, lty=3, pwd=2)
* [https://stackoverflow.com/a/12135122 Formatting Decimal places in R, round()]. [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/format format()] where '''nsmall''' controls the minimum number of digits to the right of the decimal point
</pre>
* The default digits 7 may be too small. The acceptable range is 1-22. See the following examples


In R,
== Histogram with density line ==
<syntaxhighlight lang='rsplus'>
<pre>
> options()$digits # Default
hist(x, prob = TRUE)
[1] 7
lines(density(x), col = 4, lwd = 2)
> 100000.07 + .04
</pre>
[1] 100000.1
The overlayed density may looks strange in cases for example counts from single-cell RNASeq or p-values from RNASeq (there is a peak around x=0).
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11
</syntaxhighlight>


In Python,
== Graphical Parameters, Axes and Text, Combining Plots ==
<syntaxhighlight lang='python'>
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
>>> 100000.07 + .04
100000.11
</syntaxhighlight>


== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ==
== 15 Questions All R Users Have About Plots ==
<syntaxhighlight lang='rsplus'>
See [https://www.datacamp.com/tutorial/15-questions-about-r-plots 15 Questions All R Users Have About Plots]. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)  
[1] 3.164561e-01 2.970700e+04 9.387400e+04


# Method 1. Without changing the global option
# How To Draw An Empty R Plot? plot.new()
> format(c(numer/denom, numer, denom), scientific=FALSE)
# How To Set The Axis Labels And Title Of The R Plots?
[1] "    0.3164561" "29707.0000000" "93874.0000000"
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par].
# How To Add Or Change The R Plot’s Legend? legend()
# How To Draw A Grid In Your R Plot? [https://r-charts.com/base-r/grid/ grid()]
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
# How To Adjust The Size Of Points In An R Plot? cex argument
# How To Fit A Smooth Curve To Your R Data? loess() and lines()
# How To Add Error Bars In An R Plot? arrows()
# How To Save A Plot As An Image On Disc
# How To Plot Two R Plots Next To Each Other? '''par(mfrow)'''[which means Multiple Figures (use ROW-wise)], '''gridBase''' package, '''lattice''' package
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
# What Is The Function Of hjust And vjust In ggplot2?


# Method 2. Change the global option
== jitter function ==
> options(scipen=999)
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
> numer/denom
** jitter(, amount) function adds a random variation between -amount/2 and amount/2 to each element in x
[1] 0.3164561
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
> c(numer/denom, numer, denom)
* [https://www.r-bloggers.com/2023/09/when-to-use-jitter/ When to use Jitter]
[1]    0.3164561 29707.0000000 93874.0000000
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]
> c(4/5, numer, denom)
[1]    0.8 29707.0 93874.0
</syntaxhighlight>


== sprintf ==
:[[File:Jitterbox.png|200px]]
=== Format number as fixed width, with leading zeros ===
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1


<syntaxhighlight lang='rsplus'>
== Scatterplot with the "rug" function ==
# sprintf()
<pre>
a <- seq(1,101,25)
require(stats) # both 'density' and its default method
sprintf("name_%03d", a)
with(faithful, {
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
    plot(density(eruptions, bw = 0.15))
    rug(eruptions)
    rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
})
</pre>
[[:File:RugFunction.png]]
 
See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.


# formatC()
== Identify/Locate Points in a Scatter Plot ==
paste("name", formatC(a, width=3, flag="0"), sep="_")
<ul>
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
<li>[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
</syntaxhighlight>
<li>[https://stackoverflow.com/a/23234142 Using the identify function in R]
<pre>
plot(x, y)
identify(x, y, labels = names, plot = TRUE)  
# Use left clicks to select points we want to identify and "esc" to stop the process
# This will put the labels on the plot and also return the indices of points
# [1] 143
names[143]
</pre>
</ul>


=== sprintf does not print ===
== Draw a single plot with two different y-axes ==
Use cat() or print() outside sprintf(). sprintf() do not print in a non interactive mode.
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html
<syntaxhighlight lang='rsplus'>
cat(sprintf('%5.2f\t%i\n',1.234, l234))
</syntaxhighlight>


== Creating publication quality graphs in R ==
== Draw Color Palette ==
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/


== HDF5 : Hierarchical Data Format==
=== Default palette before R 4.0 ===
HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.
palette() # black, red, green3, blue, cyan, magenta, yellow, gray


* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
<pre>
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
* rhdf5 is used by [http://amp.pharm.mssm.edu/archs4/data.html ARCHS4] where you can download R program that will download hdf5 file storing expression and metadata such as gene ID, sample/GSM ID, tissues, et al.
palette(tropical)
plot(1:5, 1:5, col=1:5, pch=16, cex=5)
</pre>


<syntaxhighlight lang='rsplus'>
=== New palette in R 4.0.0 ===
> h5ls(destination_file)
[https://youtu.be/I4k0LkTOKvU?t=464 R 4.0: 3 new features], [https://blog.revolutionanalytics.com/2020/04/r-400-is-released.html R 4.0.0 now available, and a look back at R's history]. For example, we can select "ggplot2" palette to make the base graphics charts that match the color scheme of ggplot2.
  group                          name      otype  dclass          dim
<pre>
0     /                          data  H5I_GROUP                     
R> palette()
/data                    expression H5I_DATASET INTEGER 35238 x 65429
[1] "black"  "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
2      /                           info  H5I_GROUP                     
[8] "gray62"
/info                        author H5I_DATASET  STRING            1
R> palette.pals()
4 /info                        contact H5I_DATASET  STRING            1
  [1] "R3"              "R4"              "ggplot2"       
/info                  creation-date H5I_DATASET  STRING            1
  [4] "Okabe-Ito"       "Accent"          "Dark 2"          
/info                            lab H5I_DATASET  STRING            1
  [7] "Paired"         "Pastel 1"       "Pastel 2"        
/info                        version H5I_DATASET  STRING            1
[10] "Set 1"          "Set 2"          "Set 3"          
8      /                           meta  H5I_GROUP                     
[13] "Tableau 10"      "Classic Tableau" "Polychrome 36"  
9  /meta          Sample_channel_count H5I_DATASET  STRING        65429
[16] "Alphabet"
10 /meta    Sample_characteristics_ch1 H5I_DATASET  STRING        65429
R> palette.colors(palette='R4') # same as palette()
11 /meta        Sample_contact_address H5I_DATASET  STRING        65429
[1] "#000000" "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
12 /meta            Sample_contact_city H5I_DATASET  STRING        65429
[8] "#9E9E9E"
13 /meta        Sample_contact_country H5I_DATASET  STRING        65429
R> palette("R3") # nothing return on screen but palette has changed
14 /meta      Sample_contact_department H5I_DATASET STRING        65429
R> palette()
15 /meta          Sample_contact_email H5I_DATASET STRING        65429
[1] "black"  "red"    "green3" "blue"   "cyan"   "magenta" "yellow"
16 /meta       Sample_contact_institute H5I_DATASET  STRING         65429
[8] "gray"  
17 /meta      Sample_contact_laboratory H5I_DATASET STRING        65429
R> palette("R4") # reset to the default color palette; OR palette("default")
18 /meta            Sample_contact_name H5I_DATASET  STRING        65429
19 /meta          Sample_contact_phone H5I_DATASET  STRING        65429
20 /meta Sample_contact_zip-postal_code H5I_DATASET  STRING        65429
21 /meta        Sample_data_processing H5I_DATASET  STRING        65429
22 /meta         Sample_data_row_count H5I_DATASET  STRING        65429
23 /meta            Sample_description H5I_DATASET  STRING        65429
24 /meta    Sample_extract_protocol_ch1 H5I_DATASET  STRING        65429
25 /meta          Sample_geo_accession H5I_DATASET  STRING        65429
26 /meta       Sample_instrument_model H5I_DATASET  STRING        65429
27 /meta        Sample_last_update_date H5I_DATASET  STRING        65429
28 /meta       Sample_library_selection H5I_DATASET  STRING        65429
29 /meta         Sample_library_source H5I_DATASET  STRING        65429
30 /meta        Sample_library_strategy H5I_DATASET STRING        65429
31 /meta            Sample_molecule_ch1 H5I_DATASET  STRING        65429
32 /meta            Sample_organism_ch1 H5I_DATASET  STRING        65429
33 /meta            Sample_platform_id H5I_DATASET  STRING        65429
34 /meta                Sample_relation H5I_DATASET  STRING        65429
35 /meta              Sample_series_id H5I_DATASET STRING        65429
36 /meta        Sample_source_name_ch1 H5I_DATASET  STRING        65429
37 /meta                  Sample_status H5I_DATASET STRING        65429
38 /meta        Sample_submission_date H5I_DATASET  STRING        65429
39 /meta   Sample_supplementary_file_1 H5I_DATASET  STRING        65429
40 /meta   Sample_supplementary_file_2 H5I_DATASET  STRING        65429
41 /meta              Sample_taxid_ch1 H5I_DATASET  STRING        65429
42 /meta                  Sample_title H5I_DATASET  STRING        65429
43 /meta                    Sample_type H5I_DATASET  STRING        65429
44 /meta                          genes H5I_DATASET STRING        35238
</syntaxhighlight>


== Formats for writing/saving and sharing data ==
R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
[http://www.econometricsbysimulation.com/2016/12/efficiently-saving-and-sharing-data-in-r_46.html Efficiently Saving and Sharing Data in R]
R> for(id in palette.pals()) {
    scales::show_col(palette.colors(palette = id))
    title(id)
    readline("Press [enter] to continue")
  }
</pre>
The '''palette''' function can also be used to change the color palette. See [https://data.library.virginia.edu/setting-up-color-palettes-in-r/ Setting up Color Palettes in R]
<pre>
palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
  # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))


== Write unix format files on Windows and vice versa ==
cc <- palette()
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html
palette(c(cc,"purple","brown")) # Add two colors
</pre>
<pre>
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502
</pre>


== with() and within() functions ==
=== evoPalette ===
within() is similar to with() except it is used to create new columns and merge them with the original data sets. See [http://www.youtube.com/watch?v=pZ6Bnxg9E8w&list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP youtube video].
[http://gradientdescending.com/evolve-new-colour-palettes-in-r-with-evopalette/ Evolve new colour palettes in R with evoPalette]
<pre>
closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)


mk <- within(mariokart, {
=== rtist ===
            closePr <- totalPr - shipPr
[https://github.com/tomasokal/rtist?s=09 rtist]: Use the palettes of famous artists in your own visualizations.
    })
head(mk) # new column closePr


mk <- mariokart
== SVG ==
aggregate(. ~ wheels + cond, mk, mean)
=== Embed svg in html ===
# create mean according to each level of (wheels, cond)
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html


aggregate(totalPr ~ wheels + cond, mk, mean)
=== svglite ===
svglite is better R's svg(). It was used by ggsave().
[https://www.rstudio.com/blog/svglite-1-2-0/ svglite 1.2.0], [https://r-graphics.org/recipe-output-vector-svg R Graphics Cookbook].


tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)
=== pdf -> svg ===
</pre>
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].


== stem(): stem-and-leaf plot, bar chart on terminals ==
=== svg -> png ===
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
[https://laustep.github.io/stlahblog/posts/SVG2PNG.html SVG to PNG] using the [https://cran.rstudio.com/web/packages/gyro/index.html gyro] package
* https://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package


== Graphical Parameters, Axes and Text, Combining Plots ==
== read.table ==
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
=== clipboard ===
{{Pre}}
source("clipboard")
read.table("clipboard")
</pre>


== 15 Questions All R Users Have About Plots ==
=== inline text ===
See http://blog.datacamp.com/15-questions-about-r-plots/. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.
{{Pre}}
mydf <- read.table(header=T, text='
cond yval
    A 2
    B 2.5
    C 1.6
')
</pre>


# How To Draw An Empty R Plot? plot.new()
=== http(s) connection ===
# How To Set The Axis Labels And Title Of The R Plots?
{{Pre}}
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext()
                          ssl.verifypeer = FALSE)
# How To Add Or Change The R Plot’s Legend? legend()
ip <- read.table(textConnection(temp), as.is=TRUE)
# How To Draw A Grid In Your R Plot? grid()
</pre>
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
# How To Adjust The Size Of Points In An R Plot? cex argument
# How To Fit A Smooth Curve To Your R Data? loess() and lines()
# How To Add Error Bars In An R Plot? arrows()
# How To Save A Plot As An Image On Disc
# How To Plot Two R Plots Next To Each Other? par(mfrow), '''gridBase''' package, '''lattice''' package
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
# What Is The Function Of hjust And vjust In ggplot2?


== jitter function ==
=== read only specific columns ===
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.
* [https://statistical-programming.com/jitter-r-function-example/ The jitter R Function] 3 Example Codes (Basic Application & Boxplot Visualization)
{{Pre}}
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
#
system.time(x <- read.delim("Methylation450k.txt",  
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
</pre>


== Scatterplot with the "rug" function ==
To know the number of columns, we might want to read the first row first.
<pre>
{{Pre}}
require(stats) # both 'density' and its default method
library(magrittr)
with(faithful, {
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
    plot(density(eruptions, bw = 0.15))
    rug(eruptions)
    rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
})
</pre>
</pre>
[[File:RugFunction.png|200px]]


See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.
Another method is to use '''pipe()''', '''cut''' or '''awk'''. See [https://stackoverflow.com/questions/2193742/ways-to-read-only-select-columns-from-a-file-into-r-a-happy-medium-between-re ways to read only selected columns from a file into R]


== Identify/Locate Points in a Scatter Plot ==
=== check.names = FALSE in read.table() ===
[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
<pre>
gx <- read.table(file, header = T, row.names =1)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH."        "IGHV1.69"    "IGKV1.5"    "IGKV2.24"    "KRTAP13.2" 
# [7] "KRTAP19.1"  "KRTAP2.4"    "KRTAP5.9"    "KRTAP6.3"    "Kua.UEV" 


== Draw a single plot with two different y-axes ==
gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH@"        "IGHV1-69"    "IGKV1-5"    "IGKV2-24"    "KRTAP13-2" 
# [7] "KRTAP19-1"  "KRTAP2-4"    "KRTAP5-9"    "KRTAP6-3"    "Kua-UEV" 
</pre>


== Draw Color Palette ==
=== setNames() ===
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
Change the colnames. See an example from [https://www.tidymodels.org/start/models/ tidymodels]


== SVG ==
=== Testing for valid variable names ===
=== Embed svg in html ===
[https://www.r-bloggers.com/testing-for-valid-variable-names/ Testing for valid variable names]
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html


=== svglite ===
=== make.names(): Make syntactically valid names out of character vectors ===
https://blog.rstudio.org/2016/11/14/svglite-1-2-0/
* [https://stat.ethz.ch/R-manual/R-devel/library/base/html/make.names.html make.names()]
* A valid variable name consists of letters, numbers and the '''dot''' or '''underline''' characters. The variable name starts with a letter or the dot not followed by a number. See [https://www.tutorialspoint.com/r/r_variables.htm R variables].
<pre>
make.names("abc-d") # [1] "abc.d"
</pre>


=== pdf -> svg ===
== Serialization ==
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the
[https://stat.ethz.ch/pipermail/r-devel/attachments/20130628/56473803/attachment.pl post] on R mailing list.
<pre>
> a <- list(1,2,3)
> a_serial <- serialize(a, NULL)
> a_length <- length(a_serial)
> a_length
[1] 70
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)
</pre>
In C++ process, I receive one int variable first to get the length, and
then read <length> bytes from the connection.


== read.table ==
== socketConnection ==
=== clipboard ===
See ?socketconnection.  
<syntaxhighlight lang="rsplus">
source("clipboard")
read.table("clipboard")
</syntaxhighlight>


=== inline text ===
=== Simple example ===
<syntaxhighlight lang="rsplus">
from the socketConnection's manual.
mydf <- read.table(header=T, text='
cond yval
    A 2
    B 2.5
    C 1.6
')
</syntaxhighlight>


=== http(s) connection ===
Open one R session
<syntaxhighlight lang="rsplus">
<pre>
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
                          ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
</syntaxhighlight>
 
=== read only specific columns ===
Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.
<syntaxhighlight lang="rsplus">
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
#
system.time(x <- read.delim("Methylation450k.txt",
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
</syntaxhighlight>
 
To know the number of columns, we might want to read the first row first.
<syntaxhighlight lang="rsplus">
library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</syntaxhighlight>
 
Another method is to use '''pipe()''', '''cut''' or '''awk'''. See [https://stackoverflow.com/questions/2193742/ways-to-read-only-select-columns-from-a-file-into-r-a-happy-medium-between-re ways to read only selected columns from a file into R]
 
== Serialization ==
If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the
[https://stat.ethz.ch/pipermail/r-devel/attachments/20130628/56473803/attachment.pl post] on R mailing list.
<pre>
> a <- list(1,2,3)
> a_serial <- serialize(a, NULL)
> a_length <- length(a_serial)
> a_length
[1] 70
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)
</pre>
In C++ process, I receive one int variable first to get the length, and
then read <length> bytes from the connection.
 
== socketConnection ==
See ?socketconnection.
 
=== Simple example ===
from the socketConnection's manual.
 
Open one R session
<pre>
con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
writeLines(LETTERS, con1)
close(con1)
close(con1)
Line 5,153: Line 5,787:
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]


== as.formula() ==
=== Careful on NA value ===
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<pre>
R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3  3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6
</pre>
 
=== Implicit looping ===
<pre>
set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]    # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7
</pre>
 
== modelling ==
=== update() ===
* [https://www.rdocumentation.org/packages/stats/versions/3.6.1/topics/update ?update]
* [https://stackoverflow.com/a/5118337 Reusing a Model Built in R]
 
=== Extract all variable names in lm(), glm(), ... ===
all.vars(formula(Model)[-2])
 
=== as.formula(): use a string in formula in lm(), glm(), ... ===
* [https://www.r-bloggers.com/2019/08/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
<syntaxhighlight lang='rsplus'>
{{Pre}}
? as.formula
? as.formula
xnam <- paste("x", 1:25, sep="")
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</syntaxhighlight>
</pre>
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
<syntaxhighlight lang='rsplus'>
{{Pre}}
outcome <- "mpg"
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")
variables <- c("cyl", "disp", "hp", "carb")
Line 5,209: Line 5,883:
(Intercept)          cyl        disp          hp        carb   
(Intercept)          cyl        disp          hp        carb   
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
</syntaxhighlight>
</pre>
* [https://statisticaloddsandends.wordpress.com/2019/08/24/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula] 1. as.formula() 2. subset by [[i]] 3. get() 4. eval(parse()).
 
=== reformulate ===
[https://www.r-bloggers.com/2023/06/simplifying-model-formulas-with-the-r-function-reformulate/ Simplifying Model Formulas with the R Function ‘reformulate()’]
 
=== I() function ===
I() means isolates. See [https://stackoverflow.com/a/24192745 What does the capital letter "I" in R linear regression formula mean?],  [https://stackoverflow.com/a/8055683 In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)]
 
=== Aggregating results from linear model ===
https://stats.stackexchange.com/a/6862
 
== Replacement function "fun(x) <- a" ==
[https://stackoverflow.com/questions/11563154/what-are-replacement-functions-in-r What are Replacement Functions in R?]
<pre>
R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
    x[x > value] <- Inf
    x
}
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]  1  3 Inf Inf
 
R> "cutoff<-"(x = xx, value = 65)
[1]  1  3 Inf Inf
</pre>
The statement '''fun(x) <- a''' and R will read '''x <- "fun<-"(x,a) '''


== S3 and S4 methods ==
== S3 and S4 methods and signature ==
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* Software for Data Analysis: Programming with R by John Chambers
* Software for Data Analysis: Programming with R by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* [https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers, 2016
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods]
* [https://rstudio-education.github.io/hopr/s3.html Hands-On Programming with R] by Garrett Grolemund
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package]
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package], * [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://adv-r.had.co.nz/S4.html, http://adv-r.had.co.nz/OO-essentials.html
* http://adv-r.had.co.nz/S4.html
* [https://appsilon.com/object-oriented-programming-in-r-part-1/ Object-Oriented Programming in R (Part 1): An Introduction], [https://appsilon.com/object-oriented-programming-in-r-part-2/ Part 2: S3 Simplified]


To get the source code of S4 methods, we can use showMethod(), getMethod() and showMethod(). For example
=== Debug an S4 function ===
<syntaxhighlight lang='rsplus'>
* '''showMethods('FUNCTION')'''
library(qrqc)
* '''getMethod('FUNCTION', 'SIGNATURE') ''' 
showMethods("gcPlot")
* '''debug(, signature)'''
getMethod("gcPlot", "FASTQSummary") # get an error
{{Pre}}
showMethods("gcPlot", "FASTQSummary") # good.
> args(debug)
</syntaxhighlight>
function (fun, text = "", condition = NULL, signature = NULL)  


* '''Debug a S4 function'''
<syntaxhighlight lang='rsplus'>
> library(genefilter) # Bioconductor
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
> showMethods("nsFilter")
Line 5,239: Line 5,941:
eset="ExpressionSet"
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")
> debug(nsFilter, signature="ExpressionSet")
</syntaxhighlight>
 
library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code
</pre>
See the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L367 source code] of '''normalizationFactors<-''' (setReplaceMethod() is used) and the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L385 source code] of '''estimateSizeFactors()'''. We can see how ''avgTxLength'' was used in estimateNormFactors().
 
Another example
<pre>
library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)
 
showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"
 
debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz,
#      normalization = ssgsea.norm, verbose = verbose,
#      BPPARAM = BPPARAM))
 
isdebugged("gsva")
# [1] TRUE
undebug(gsva)
</pre>


* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
<syntaxhighlight lang='rsplus'>
{{Pre}}
library(IRanges)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir <- IRanges(start=c(10, 20, 30), width=5)
Line 5,273: Line 6,009:
##  
##  
## Known Subclasses: "NormalIRanges"
## Known Subclasses: "NormalIRanges"
</syntaxhighlight>
</pre>


=== See what methods work on an object ===
=== Check if a function is an S4 method ===
see what methods work on an object, e.g. a GRanges object:
'''isS4(foo)'''
<syntaxhighlight lang='rsplus'>methods(class="GRanges")</syntaxhighlight> Or if you have an object, x: <syntaxhighlight lang='rsplus'>methods(class=class(x))</syntaxhighlight>


=== View S3 function definition: double colon '::' and triple colon ':::' operators ===
=== How to access the slots of an S4 object ===
?":::"
* @ will let you access the slots of an S4 object.
* Note that often the best way to do this is to not access the slot directly but rather through an accessor function (e.g. coefs() rather than digging out the coefficients with $ or @). However, often such functions do not exist so you have to access the slots directly. This will mean that your code breaks if the internal implementation changes, however.
* [https://kasperdanielhansen.github.io/genbioconductor/html/R_S4.html#slots-and-accessor-functions R - S4 Classes and Methods] Hansen. '''getClass()''' or '''getClassDef()'''.


* pkg::name returns the value of the exported variable name in namespace pkg
=== setReplaceMethod() ===
* pkg:::name returns the value of the internal variable name
* [https://stackoverflow.com/a/24253311 What's the difference between setMethod(“$<-”) and set setReplaceMethod(“$”)?]
* [https://stackoverflow.com/a/49267668 What is setReplaceMethod() and how does it work?]


<syntaxhighlight lang='rsplus'>
=== See what methods work on an object ===
see what methods work on an object, e.g. a GRanges object:
<pre>
methods(class="GRanges")
</pre>
Or if you have an object, x:
<pre>
methods(class=class(x))
</pre>
 
=== View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere() ===
?":::"
 
* pkg::name returns the value of the exported variable name in namespace pkg
* pkg:::name returns the value of the internal variable name
 
<pre>
base::"+"
base::"+"
stats:::coef.default
stats:::coef.default
</syntaxhighlight>
 
predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR 
getS3method("predict", "ppr")
 
getS3method("t", "test")
</pre>
 
[https://stackoverflow.com/a/19226817 methods() + getAnywhere() functions]


=== Read the source code (include Fortran/C, S3 and S4 methods) ===
=== Read the source code (include Fortran/C, S3 and S4 methods) ===
* [https://github.com/jimhester/lookup#readme lookup] package
* [https://github.com/jimhester/lookup#readme lookup] package
* [https://blog.r-hub.io/2019/05/14/read-the-source/ Read the source]
* [https://blog.r-hub.io/2019/05/14/read-the-source/ Read the source]
* Find the source code in [https://stackoverflow.com/a/19226817 UseMethod("XXX")] for S3 methods.
=== S3 method is overwritten ===
For example, the select() method from dplyr is overwritten by [https://github.com/cran/grpreg/blob/master/NAMESPACE grpreg] package.
An easy solution is to load grpreg before loading dplyr.
* https://stackoverflow.com/a/14407095
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods] and [https://github.com/njtierney/A-Simple-Guide-to-S3-Methods/blob/master/SimpleS3.Rmd its source]
* [https://developer.r-project.org/Blog/public/2019/08/19/s3-method-lookup/index.html S3 Method Lookup]


=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
Line 5,300: Line 6,075:


For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
<syntaxhighlight lang='rsplus'>
{{Pre}}
> mcols(ddsNoPrior[genes, ])
> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
DataFrame with 2 rows and 21 columns
Line 5,319: Line 6,094:
1      TRUE        3  210.4045 0.2648753
1      TRUE        3  210.4045 0.2648753
2      TRUE        9  243.7455 0.3248949
2      TRUE        9  243.7455 0.3248949
</syntaxhighlight>
</pre>
 
== Pipe ==
* [https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
* [https://win-vector.com/2020/12/05/r-is-getting-an-official-pipe-operator/ R is Getting an Official Pipe Operator], [https://win-vector.com/2020/12/07/my-opinion-on-rs-upcoming-pipe/ My Opinion on R’s Upcoming Pipe]
<ul>
<li> a(b(x)) vs '''x |> b() |> a()'''. See [https://twitter.com/henrikbengtsson/status/1335328090390597632 this tweet] in R-dev 2020-12-04.
<pre>
e0 <- quote(a(b(x)))
e1 <- quote(x |> b() |> a())
identical(e0, e1)
</pre>
</li>
<li>
[https://selbydavid.com/2021/05/18/pipes/ There are now 3 different R pipes]
</li>
<li>[https://stackoverflow.com/a/67629310 Error: The pipe operator requires a function call as RHS].
<pre>
# native pipe
foo |> bar()
# magrittr pipe
foo %>% bar
</pre>
</li>
<li>[https://www.infoworld.com/article/3621369/use-the-new-r-pipe-built-into-r-41.html Use the new R pipe built into R 4.1] </li>
<li>[https://towardsdatascience.com/the-new-native-pipe-operator-in-r-cbc5fa8a37bd The New Native Pipe Operator in R] </li>
<li>[https://ivelasq.rbind.io/blog/understanding-the-r-pipe/ Understanding the native R pipe |> ] </li>
</ul>
 
Packages take advantage of pipes
<ul>
<li>[https://cran.r-project.org/web/packages/rstatix/index.html rstatix]: Pipe-Friendly Framework for Basic Statistical Tests
</ul>


== findInterval() ==
== findInterval() ==
Line 5,325: Line 6,132:
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
== Assign operator ==
* Earlier versions of R used underscore (_) as an assignment operator.
* [https://developer.r-project.org/equalAssign.html Assignments with the = Operator]
* In R 1.8.0 (2003), the assign operator has been removed. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
* In R 1.9.0 (2004), "_" is allowed in valid names. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
: [[File:R162.png|200px]]
== Operator precedence ==
The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.


== order(), rank() and sort() ==
== order(), rank() and sort() ==
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
<syntaxhighlight lang='rsplus'>
<pre>
> x = sample(10)
> x = sample(10)
> x
> x
Line 5,343: Line 6,161:
> sort(x)
> sort(x)
  [1]  1  2  3  4  5  6  7  8  9 10
  [1]  1  2  3  4  5  6  7  8  9 10
</syntaxhighlight>
</pre>
 
=== OS-dependent results on sorting string vector ===
Gene symbol case.
<pre>
# mac:
order(c("DC-UbP", "DC2")) # c(1,2)
 
# linux:
order(c("DC-UbP", "DC2")) # c(2,1)
</pre>


== do.call, rbind, lapply ==
Affymetric id case.
Lots of examples. See for example [https://stat.ethz.ch/pipermail/r-help/attachments/20140423/62d8d103/attachment.pl this one] for creating a data frame from a vector.
<pre>
<syntaxhighlight lang='rsplus'>
# mac:
x <- readLines(textConnection("---CLUSTER 1 ---
order(c("202800_at", "2028_s_at")) # [1] 2 1
3
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"
4
 
5
# linux
6
order(c("202800_at", "2028_s_at")) # [1] 1 2
---CLUSTER 2 ---
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"
9
</pre>
10
It does not matter if we include factor() on the character vector.
8
11"))


# create a list of where the 'clusters' are
The difference is related to locale. See
clust <- c(grep("CLUSTER", x), length(x) + 1L)


# get size of each cluster
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/locales ?locales] in R
clustSize <- diff(clust) - 1L
* On OS, type '''locale'''
* [https://stackoverflow.com/questions/39171613/sort-produces-different-results-in-ubuntu-and-windows sort() produces different results in Ubuntu and Windows]
* To fix the inconsistency problem, we can set the locale in R code to "C" or use the stringr package (the locale is part of [https://www.rdocumentation.org/packages/stringr/versions/1.4.0/topics/str_order str_order()]'s arguments).
<pre>
# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2


# get cluster number
# Or setting the locale to "C"
clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])
Sys.setlocale("LC_ALL", "C"); sort(c("DC-UbP", "DC2"))
# Or
Sys.setlocale("LC_COLLATE", "C"); sort(c("DC-UbP", "DC2"))
# But not
Sys.setlocale("LC_ALL", "en_US.UTF-8"); sort(c("DC-UbP", "DC2"))
</pre>


result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
=== unique() ===
    cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
It seems it does not sort. [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/unique ?unique].
        , Cluster = .cl
<pre>
        )
# mac & linux
    }))
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"
</pre>


result
== do.call ==
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.


    Object Cluster
[https://www.r-bloggers.com/2023/05/the-do-call-function-in-r-unlocking-efficiency-and-flexibility/ The do.call() function in R: Unlocking Efficiency and Flexibility]
[1,] "3"    "1"
[2,] "4"    "1"
[3,] "5"    "1"
[4,] "6"    "1"
[5,] "9"    "2"
[6,] "10"  "2"
[7,] "8"    "2"
[8,] "11"  "2"
</syntaxhighlight>


A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html help].


Another example is to reproduce aggregate(). aggregate() = do.call() + by().
* Usage
<syntaxhighlight lang='rsplus'>
{{Pre}}
attach(mtcars)
do.call(what, args, quote = FALSE, envir = parent.frame())
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# what: either a function or a non-empty character string naming the function to be called.
# the above approach give the same result as the following
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# except it does not have an extra Group.x columns
# quote: a logical value indicating whether to quote the arguments.
aggregate(mtcars, list(cyl, vs), FUN=mean)
# envir: an environment within which to evaluate the call. This will be most useful
</syntaxhighlight>
#        if what is a character string and the arguments are symbols or quoted expressions.
 
</pre>
== How to get examples from help file ==
* do.call() is similar to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/lapply lapply()] but not the same. It seems do.call() can make a simple function vectorized.
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
{{Pre}}
Method 1:
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
> lapply(list(imag = 1:3), complex)
$imag
[1] 0+0i
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
> do.call(function(x) x+1, list(1:3))
[1] 2 3 4
</pre>
* Applying do.call with Multiple Arguments
<pre>
<pre>
example(acf, give.lines=TRUE)
> do.call("sum", list(c(1,2,3,NA), na.rm = TRUE))
[1] 6
> do.call("sum", list(c(1,2,3,NA) ))
[1] NA
</pre>
</pre>
Method 2:
* [https://www.stat.berkeley.edu/~s133/Docall.html do.call() allows you to call any R function, but instead of writing out the arguments one by one, you can use a list to hold the arguments of the function.]
<pre>
{{Pre}}
Rd <- utils:::.getHelpFile(?acf)
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
tools::Rd2ex(Rd)
> length(tmp)
[1] 3
> tmp[1:4,]
  Var1 Var2 Var3
1    a    1    +
2    b    1    +
3    a    2    +
4    b    2    +
> c(tmp, sep = "")
$Var1
[1] a b a b a b a b a b a b
Levels: a b
 
$Var2
[1] 1 1 2 2 3 3 1 1 2 2 3 3
 
$Var3
[1] + + + + + + - - - - - -
Levels: + -
 
$sep
[1] ""
> do.call("paste", c(tmp, sep = ""))
[1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[12] "b3-"
</pre>
</pre>
* ''environment'' and ''quote'' arguments.
{{Pre}}
> A <- 2
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A) 
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env) 
[1] 4
> do.call(f, list(A), envir = env) 
[1] 2                      # Why?


== "[" and "[[" with the sapply() function ==
> eval(call("f", A))                     
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
[1] 2
> eval(call("f", quote(A)))             
[1] 2
> eval(call("f", A), envir = env)       
[1] 4
> eval(call("f", quote(A)), envir = env) 
[1] 100
</pre>
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
{{Pre}}
> foo <- function(a=1, b=2, ...) {
        list(arg=do.call(c, as.list(match.call())[-1]))
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a
1
> foo(a=1, b=2, c=3)
$arg
a b c
1 2 3
</pre>
* do.call() + switch(). See [https://github.com/satijalab/seurat/blob/13b615c27eeeac85e5c928aa752197ac224339b9/R/preprocessing.R#L2450 an example] from Seurat::NormalizeData.
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
do.call(
  what = switch(
    EXPR = margin,
    '1' = 'rbind',
    '2' = 'cbind',
    stop("'margin' must be 1 or 2")
  ),
  args = normalized.data
)
switch('a', 'a' = rnorm(3), 'b'=rnorm(4)) # switch returns a value
do.call(switch('a', 'a' = 'rnorm', 'b'='rexp'), args=list(n=4)) # switch returns a function
</pre>
</pre>
is the same as
* The function we want to call is a string that may change: [https://github.com/cran/glmnet/blob/master/R/cv.glmnet.raw.R#L66 glmnet]
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
# Suppose we want to call cv.glmnet or cv.coxnet or cv.lognet or cv.elnet .... depending on the case
fun = paste("cv", subclass, sep = ".")
cvstuff = do.call(fun, list(predmat,y,type.measure,weights,foldid,grouped))
</pre>
</pre>


== Dealing with date ==
=== expand.grid, mapply, vapply ===
<pre>
[https://shikokuchuo.net/posts/10-combinations/ A faster way to generate combinations for mapply and vapply]
d1 = date()
class(d1) # "character"
d2 = Sys.Date()
class(d2) # "Date"


format(d2, "%a %b %d")
=== do.call vs mapply ===
* do.call() is doing what [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/mapply mapply()] does but do.call() uses a list instead of multiple arguments. So do.call() more close to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/funprog base::Map()] function.
{{Pre}}
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
      Var1
[1,] "a1+"
[2,] "b1+"
[3,] "a2+"
[4,] "b2+"
[5,] "a3+"
[6,] "b3+"
[7,] "a1-"
[8,] "b1-"
[9,] "a2-"
[10,] "b2-"
[11,] "a3-"
[12,] "b3-"
# It does not work if we do not explicitly specify the arguments in mapply()
> mapply(paste, tmp, sep = "")
      Var1 Var2 Var3
[1,] "a"  "1"  "+"
[2,] "b"  "1"  "+"
[3,] "a"  "2"  "+"
[4,] "b"  "2"  "+"
[5,] "a"  "3"  "+"
[6,] "b" "3"  "+"
[7,] "a"  "1"  "-"
[8,] "b"  "1"  "-"
[9,] "a"  "2"  "-"
[10,] "b"  "2"  "-"
[11,] "a"  "3"  "-"
[12,] "b"  "3"  "-"
</pre>
* mapply is useful in generating variables with a vector of parameters. For example suppose we want to generate variables from exponential/weibull distribution and a vector of scale parameters (depending on some covariates). In this case we can use ([https://stackoverflow.com/a/17031993 Simulating Weibull distributions from vectors of parameters in R])
{{Pre}}
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
# [1] 1.326108 9.885284
set.seed(1)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
dim(x) # [1]  2 1000
rowMeans(x)
# [1]  1.032209 10.104131
</pre>
{{Pre}}
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))
</pre>


library(lubridate); ymd("20140108") # "2014-01-08 UTC"
=== do.call vs lapply ===
mdy("08/04/2013") # "2013-08-04 UTC"
[https://stackoverflow.com/a/10801883 What's the difference between lapply and do.call?] It seems to me the best usage is combining both functions: '''do.call(..., lapply())'''
dmy("03-04-2013") # "2013-04-03 UTC"
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland")
# "2011-08-03 10:15:03 NZST"
?Sys.timezone
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues
</pre>
* http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
* http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
* We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.


== [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] and deparse/substitute ==
* lapply returns a list of the same length as X, each element of which is the result of applying FUN to the corresponding element of X.
* [https://cran.r-project.org/web/packages/lazyeval/vignettes/lazyeval.html Vignette] from the [https://cran.r-project.org/web/packages/lazyeval/index.html lazyeval] package. It is needed in three cases
* do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it. '''It is widely used, for example, to assemble lists into simpler structures (often with rbind or cbind).'''
** Labelling: turn an argument into a label
* Map applies a function to the corresponding elements of given vectors... Map is a simple wrapper to mapply which does not attempt to simplify the result, similar to Common Lisp's mapcar (with arguments being recycled, however). Future versions may allow some control of the result type.
** Formulas
** Dot-dot-dot
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression.
** substitute() is often paired with deparse() to create informative labels for data sets and plots.  
** Use 'substitute' to include the variable's name in a plot title, e.g.: '''var <- "abc"; hist(var,main=substitute(paste("Dist of ", var))) ''' will show the title "Dist of var" instead of "Dist of abc" in the title.
* quote(expr) - similar to substitute() but do nothing?? [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/noquote noquote] - print character strings without quotes
* eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
* deparse(expr) - turns unevaluated expressions into character strings. For example,  
<pre>
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                   
[4] "NULL"   


> deparse(args(lm), width=20)
{{Pre}}
[1] "function (formula, data, "        "    subset, weights, "         
> lapply(iris, class) # same as Map(class, iris)
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, " 
$Sepal.Length
[5] "    y = FALSE, qr = TRUE, "      "    singular.ok = TRUE, "       
[1] "numeric"
[7] "   contrasts = NULL, "           "    offset, ...) "             
[9] "NULL"
</pre>
* parse(text) - returns the parsed but unevaluated expressions in a list. See [[R#Create_a_Simple_Socket_Server_in_R|Create a Simple Socket Server in R]] for the application of '''eval(parse(text))'''. Be cautious!
** [http://r.789695.n4.nabble.com/using-eval-parse-paste-in-a-loop-td849207.html eval(parse...)) should generally be avoided]
** [https://stackoverflow.com/questions/13649979/what-specifically-are-the-dangers-of-evalparse What specifically are the dangers of eval(parse(…))?]


Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).
$Sepal.Width
<syntaxhighlight lang='rsplus'>
[1] "numeric"
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3


f1(1:3)
$Petal.Length
f2(1:3)
[1] "numeric"
f3(1:3)


# Or
$Petal.Width
myfun <- function(f, a) {
[1] "numeric"
    eval(parse(text = f))(a)
 
}
$Species
myfun("f1", 1:3)
[1] "factor"
myfun("f2", 1:3)
 
myfun("f3", 1:3)
> x <- lapply(iris, class)
> do.call(c, x)
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species
  "numeric"    "numeric"    "numeric"   "numeric"     "factor"  
</pre>


# Or with lapply
https://stackoverflow.com/a/10801902
method <- c("f1", "f2", "f3")
* '''lapply''' applies a function '''over a list'''. So there will be several function calls.
res <- lapply(method, function(M) {
* '''do.call''' calls a function with '''a list of arguments''' (... argument) such as [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/c c()] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/cbind rbind()/cbind()] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/sum sum] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/order order] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/Extract "["] or paste. So there is only one function call.
                    Mres <- eval(parse(text = M))(1:3)
{{Pre}}
                    return(Mres)
> X <- list(1:3,4:6,7:9)
})
> lapply(X,mean)
names(res) <- method
[[1]]
</syntaxhighlight>
[1] 2


== The ‘…’ argument ==
[[2]]
See [http://cran.r-project.org/doc/manuals/R-intro.html#The-three-dots-argument Section 10.4 of An Introduction to R]. Especially, the expression '''list(...)''' evaluates all such arguments and returns them in a named list
[1] 5


== Lazy evaluation in R functions arguments ==
[[3]]
* http://adv-r.had.co.nz/Functions.html
[1] 8
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
> do.call(sum, X)
[1] 45
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
[1] 45
> do.call(mean, X) # Error
> do.call(rbind,X)
    [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9
> lapply(X,rbind)
[[1]]
    [,1] [,2] [,3]
[1,]    1    2    3


'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.
[[2]]
    [,1] [,2] [,3]
[1,]    4    5    6


* Example 1. By default, R function arguments are lazy.
[[3]]
<pre>
    [,1] [,2] [,3]
f <- function(x) {
[1,]    7    8    9
  999
> mapply(mean, X, trim=c(0,0.5,0.1))
}
[1] 2 5 8
f(stop("This is an error!"))
> mapply(mean, X)  
#> [1] 999
[1] 2 5 8
</pre>
</pre>
Below is a good example to show the difference of lapply() and do.call() - [https://stackoverflow.com/a/42734863 Generating Random Strings].
{{Pre}}
> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
[[1]]
[1] "Y" "D" "G" "A"


* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
[[2]]
<pre>
[1] "B" "W" "K" "N"
add <- function(x) {
 
  force(x)
> lapply(x, paste0)
  function(y) x + y
[[1]]
}
[1] "Y" "D" "G" "A"
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
</pre>


* Example 3. Default arguments are evaluated inside the function.
[[2]]
<pre>
[1] "B" "W" "K" "N"
f <- function(x = ls()) {
  a <- 1
  x
}


# ls() evaluated inside f:
> lapply(x, paste0, collapse= "")
f()
[[1]]
# [1] "a" "x"
[1] "YDGA"


# ls() evaluated in global environment:
[[2]]
f(ls())
[1] "BWKN"
# [1] "add"    "adders" "f"
</pre>


* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
> do.call(paste0, x)
<pre>
[1] "YB" "DW" "GK" "AN"
x <- NULL
if (!is.null(x) && x > 0) {
 
}
</pre>
</pre>


== Backtick sign, infix/prefix/postfix operators ==  
=== do.call + rbind + lapply ===
The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in [http://adv-r.had.co.nz/Functions.html this note].
Lots of examples. See for example [https://stat.ethz.ch/pipermail/r-help/attachments/20140423/62d8d103/attachment.pl this one] for creating a data frame from a vector.
{{Pre}}
x <- readLines(textConnection("---CLUSTER 1 ---
3
4
5
6
---CLUSTER 2 ---
9
10
8
11"))


'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
# create a list of where the 'clusters' are
<pre>
clust <- c(grep("CLUSTER", x), length(x) + 1L)
1 + 2    # infix
 
+ 1 2    # prefix
# get size of each cluster
1 2 +    # postfix
clustSize <- diff(clust) - 1L
</pre>


== List data type ==
# get cluster number
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5
</pre>


== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
* http://adv-r.had.co.nz/Exceptions-Debugging.html
    cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
* try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
        , Cluster = .cl
<pre>
        )
out <- try({
    }))
  a <- 1
  b <- "x"
  a + b
})


elements <- list(1:10, c(-1, 10), c(T, F), letters)
result
results <- lapply(elements, log)
is.error <- function(x) inherits(x, "try-error")
succeeded <- !sapply(results, is.error)
</pre>
* tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
<pre>
tryCatch(expr, ..., finally)


show_condition <- function(code) {
    Object Cluster
  tryCatch(code,
[1,] "3"    "1"
    error = function(c) "error",
[2,] "4"   "1"
    warning = function(c) "warning",
[3,] "5"    "1"
    message = function(c) "message"
[4,] "6"    "1"
   )
[5,] "9"    "2"
}
[6,] "10"2"
show_condition(stop("!"))
[7,] "8"    "2"
#> [1] "error"
[8,] "11"   "2"
show_condition(warning("?!"))
</pre>
#> [1] "warning"
 
show_condition(message("?"))
A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
#> [1] "message"
 
show_condition(10)
Another example is to reproduce aggregate(). aggregate() = do.call() + by().
#> [1] 10
{{Pre}}
attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# the above approach give the same result as the following
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)
</pre>
</pre>
Below is another snippet from available.packages() function,
 
== Run examples ==
When we call help(FUN), it shows the document in the browser. The browser will show
<pre>
<pre>
z <- tryCatch(download.file(....), error = identity)
example(FUN, package = "XXX") was run in the console
if (!inherits(z, "error")) STATEMENTS
To view output in the browser, the knitr package must be installed
</pre>
</pre>


== Using list type ==
== How to get examples from help file, example() ==
=== Avoid if-else or switch ===
[https://blog.r-hub.io/2020/01/27/examples/ Code examples in the R package manuals]:
?plot.stepfun.
<pre>
<pre>
y0 <- c(1,2,4,3)
# How to run all examples from a man page
sfun0  <- stepfun(1:3, y0, f = 0)
example(within)
sfun.2 <- stepfun(1:3, y0, f = .2)
 
sfun1  <- stepfun(1:3, y0, right = TRUE)
# How to check your examples?
devtools::run_examples()  
testthat::test_examples()
</pre>


tt <- seq(0, 3, by = 0.1)
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
op <- par(mfrow = c(2,2))
Method 1:
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
<pre>
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
example(acf, give.lines=TRUE)
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
</pre>
##-- This is  revealing :
Method 2:
plot(sfun0, verticals = FALSE,
<pre>
    main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")
Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)
</pre>


for(i in 1:3)
== "[" and "[[" with the sapply() function ==
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)
<pre>
 
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
par(op)
</pre>
is the same as
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
</pre>
</pre>
[[File:StepfunExample.svg|400px]]


== Open a new Window device ==
== Dealing with dates ==
X11() or dev.new()
* Find difference
:<syntaxhighlight lang='rsplus'>
# Convert the dates to Date objects
date1 <- as.Date("6/29/21", format="%m/%d/%y")
date2 <- as.Date("11/9/21", format="%m/%d/%y")


== par() ==
# Calculate the difference in days
?par
diff_days <- as.numeric(difftime(date2, date1, units="days")) # 133
# In months
diff_days / (365.25/12) # 4.36961 


=== text size and font on main, lab & axis ===
# OR using the lubridate package
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
library(lubridate)
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]
# Convert the dates to Date objects
date1 <- mdy("6/29/21")
date2 <- mdy("11/9/21")
interval(date1, date2) %/% months(1)
</syntaxhighlight>
* http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
:<syntaxhighlight lang='rsplus'>
d1 = date()
class(d1) # "character"
d2 = Sys.Date()
class(d2) # "Date"


Examples:
format(d2, "%a %b %d")
* cex.main=0.9
* cex.lab=0.8
* font.lab=2
* cex.axis=0.8
* font.axis=2
* col.axis="grey50"


=== layout ===
library(lubridate); ymd("20140108") # "2014-01-08 UTC"
http://datascienceplus.com/adding-text-to-r-plot/
mdy("08/04/2013") # "2013-08-04 UTC"
 
dmy("03-04-2013") # "2013-04-03 UTC"
=== reset the settings ===
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
<syntaxhighlight lang='rsplus'>
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland")
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)  
# "2011-08-03 10:15:03 NZST"
....
?Sys.timezone
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues
</syntaxhighlight>
</syntaxhighlight>
* http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
* We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.
* [https://cran.r-project.org/web/packages/anytime/index.html anytime] package
* weeks to Christmas difftime(as.Date(“2019-12-25”), Sys.Date(), units =“weeks”)
* [https://blog.rsquaredacademy.com/handling-date-and-time-in-r/ A Comprehensive Introduction to Handling Date & Time in R] 2020
* [https://www.spsanderson.com/steveondata/posts/rtip-2023-05-12/index.html Working with Dates and Times Pt 1]
** Three major functions: as.Date(), as.POSIXct(), and as.POSIXlt().
** '''POSIXct''' is a class in R that represents date-time data. The ct stands for “calendar time” and it represents the (signed) number of seconds since the beginning of 1970 as a numeric vector1.  '''It stores date time as integer.'''
** '''POSIXlt''' is a class in R that represents date-time data. It stands for “local time” and is a list with components as integer vectors, which can represent a vector of broken-down times. '''It stores date time as list:sec, min, hour, mday, mon, year, wday, yday, isdst, zone, gmtoff'''.


=== mtext (margin text) vs title ===
* [https://www.r-bloggers.com/2023/11/r-lubridate-how-to-efficiently-work-with-dates-and-times-in-r/ R lubridate: How To Efficiently Work With Dates and Times in R] 2023
* https://datascienceplus.com/adding-text-to-r-plot/
* https://datascienceplus.com/mastering-r-plot-part-2-axis/


=== mgp (axis label locations) ===
== Nonstandard/non-standard evaluation, deparse/substitute and scoping ==
# The margin line (in ‘mex’ units) for the axis title, axis labels and axis line. Note that ‘mgp[1]’ affects ‘title’ whereas ‘mgp[2:3]’ affect ‘axis’. The default is ‘c(3, 1, 0). If we like to make the axis labels closer to an axis, we can use mgp=c(2.3, 1, 0) for example.
* [https://www.brodieg.com/2020/05/05/on-nse/ Standard and Non-Standard Evaluation in R]
# http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the labels (i.e. xlab and ylab in plot), the second the tick-mark labels, and third the tick marks. The default is c(3, 1, 0).
* [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] from Advanced R book.
* [https://edwinth.github.io/blog/nse/ Non-standard evaluation, how tidy eval builds on base R]
* [https://cran.r-project.org/web/packages/lazyeval/vignettes/lazyeval.html Vignette] from the [https://cran.r-project.org/web/packages/lazyeval/index.html lazyeval] package. It is needed in three cases
** Labelling: turn an argument into a label
** Formulas
** Dot-dot-dot
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression. The return mode is a '''call'''.
** substitute() is often paired with '''deparse'''() to create informative labels for data sets and plots. The return mode of deparse() is '''character strings'''.
** Use 'substitute' to include the variable's name in a plot title, e.g.: '''var <- "abc"; hist(var,main=substitute(paste("Dist of ", var))) ''' will show the title "Dist of var" instead of "Dist of abc" in the title.
** [https://stackoverflow.com/a/34079727 Passing a variable name to a function in R]
** Example:
::<syntaxhighlight lang='rsplus'>
f <- function(x) {
  substitute(x)
}
f(1:10)
# 1:10
class(f(1:10)) # or mode()
# [1] "call"
g <- function(x) deparse(substitute(x))
g(1:10)
# [1] "1:10"
class(g(1:10)) # or mode()
# [1] "character"
</syntaxhighlight>
* quote(expr) - similar to substitute() but do nothing?? [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/noquote noquote] - print character strings without quotes
:<syntaxhighlight lang='rsplus'>
mode(quote(1:10))
# [1] "call"
</syntaxhighlight>
* eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.  
** The '''parent.frame()''' is necessary in cases like the [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/update stats::update()] function used by [https://github.com/cran/glmnet/blob/master/R/relax.glmnet.R#L66 relax.glmnet()].
** Example:
::<syntaxhighlight lang='rsplus'>
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))


=== pch ===
subset1 <- function(x, condition) {
[[File:R pch.png|250px]]
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR


([https://www.statmethods.net/advgraphs/parameters.html figure source])
subset2 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌
subset2(sample_df, a == condition) # 👍
</syntaxhighlight>
* deparse(expr) - turns unevaluated expressions into character strings. For example,
:<syntaxhighlight lang='rsplus'>
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                   
[4] "NULL"   


* Full circle: pch=16
> deparse(args(lm), width=20)
[1] "function (formula, data, "        "    subset, weights, "         
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, " 
[5] "    y = FALSE, qr = TRUE, "      "    singular.ok = TRUE, "       
[7] "    contrasts = NULL, "          "    offset, ...) "             
[9] "NULL"
</syntaxhighlight>
* parse(text) - returns the parsed but unevaluated expressions in a list. See [[R#Create_a_Simple_Socket_Server_in_R|Create a Simple Socket Server in R]] for the application of '''eval(parse(text))'''. Be cautious!
** [http://r.789695.n4.nabble.com/using-eval-parse-paste-in-a-loop-td849207.html eval(parse...)) should generally be avoided]
** [https://stackoverflow.com/questions/13649979/what-specifically-are-the-dangers-of-evalparse What specifically are the dangers of eval(parse(…))?]


=== lty (line type) ===
Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).
[[File:R lty.png|250px]]
{{Pre}}
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3


([http://www.sthda.com/english/wiki/line-types-in-r-lty figure source])
f1(1:3)
f2(1:3)
f3(1:3)


=== las (label style) ===
# Or
0: The default, parallel to the axis
myfun <- function(f, a) {
    eval(parse(text = f))(a)
}
myfun("f1", 1:3)
myfun("f2", 1:3)
myfun("f3", 1:3)


1: Always horizontal
# Or with lapply
method <- c("f1", "f2", "f3")
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method
</pre>


2: Perpendicular to the axis
=== library() accept both quoted and unquoted strings ===
[https://stackoverflow.com/a/25210607 How can library() accept both quoted and unquoted strings]. The key lines are
<pre>
  if (!character.only)
    package <- as.character(substitute(package))
</pre>


3: Always vertical
=== Lexical scoping ===
* [https://lgreski.github.io/dsdepot/2020/06/28/rObjectsSObjectsAndScoping.html R Objects, S Objects, and Lexical Scoping]
* [http://www.biostat.jhsph.edu/~rpeng/docs/R-classes-scope.pdf#page=31 Dynamic scoping vs Lexical scoping] and the example of [http://www.biostat.jhsph.edu/~rpeng/docs/R-classes-scope.pdf#page=41 optimization]
* [https://www.r-bloggers.com/2024/03/indicating-local-functions-in-r-scripts/ Indicating local functions in R scripts]


=== oma (outer margin), common title for two plots ===
== The ‘…’ argument ==
The following trick is useful when we want to draw multiple plots with a common title.
* See [http://cran.r-project.org/doc/manuals/R-intro.html#The-three-dots-argument Section 10.4 of An Introduction to R]. Especially, the expression '''list(...)''' evaluates all such arguments and returns them in a named list
* [https://statisticaloddsandends.wordpress.com/2020/11/15/some-notes-when-using-dot-dot-dot-in-r/ Some notes when using dot-dot-dot (…) in R]
* [https://stackoverflow.com/questions/26684509/how-to-check-if-any-arguments-were-passed-via-ellipsis-in-r-is-missing How to check if any arguments were passed via “…” (ellipsis) in R? Is missing(…) valid?]


<syntaxhighlight lang='rsplus'>
== Functions ==
par(mfrow=c(1,2),oma = c(0, 0, 2, 0))  # oma=c(0, 0, 0, 0) by default
* https://adv-r.hadley.nz/functions.html
plot(1:10,  main="Plot 1")
* [https://towardsdatascience.com/writing-better-r-functions-best-practices-and-tips-d48ef0691c24 Writing Better R Functions — Best Practices and Tips]. The [https://cran.r-project.org/web/packages/docstring/index.html docstring] package and "?" is interesting!
plot(1:100,  main="Plot 2")
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
</syntaxhighlight>


[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
=== Function argument ===
[https://cran.r-project.org/doc/manuals/r-release/R-lang.html#Argument-matching Argument matching] from [https://cran.r-project.org/doc/manuals/r-release/R-lang.html R Language Definition] manual.


== Non-standard fonts in postscript and pdf graphics ==
Argument matching is augmented by the functions
https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.arg match.arg],
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.call match.call]
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.fun match.fun].  


== Suppress warnings ==
Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.
<syntaxhighlight lang='rsplus'>
op <- options("warn")
options(warn = -1)
....
options(op)
 
# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)
</syntaxhighlight>


== NULL, NA, NaN, Inf ==
=== Check function arguments ===
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
[https://blog.r-hub.io/2022/03/10/input-checking/ Checking the inputs of your R functions]: '''match.arg()''' , '''stopifnot()'''


== save() vs saveRDS() ==
'''stopifnot()''': function argument sanity check
# saveRDS() can only save one R object while save() does not have this constraint.
<ul>
# saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See [http://www.fromthebottomoftheheap.net/2012/04/01/saving-and-loading-r-objects/ this post].
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]. ''stopifnot'' is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.  
<pre>
<pre>
x <- 5
stopifnot(condition1, condition2, ...)
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)
</pre>
</pre>
</li>
<li>[https://rud.is/b/2020/05/19/mining-r-4-0-0-changelog-for-nuggets-of-gold-1-stopifnot/ Mining R 4.0.0 Changelog for Nuggets of Gold] </li>
</ul>


== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
=== Lazy evaluation in R functions arguments ===
* ==: exact match
* http://adv-r.had.co.nz/Functions.html
* all.equal: compare R objects x and y testing ‘near equality’
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
* identical: The safe and reliable way to test two objects for being exactly equal.
* https://twitter.com/_wurli/status/1451459394009550850
<syntaxhighlight lang='rsplus'>
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE
</syntaxhighlight>


See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  


== testhat ==
* Example 1. By default, R function arguments are lazy.
* https://github.com/r-lib/testthat
<pre>
* [http://www.win-vector.com/blog/2019/03/unit-tests-in-r/ Unit Tests in R]
f <- function(x) {
 
  999
== tinytest ==
}
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
f(stop("This is an error!"))
#> [1] 999
</pre>


== Numerical Pitfall ==
* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
<syntaxhighlight lang='bash'>
.1 - .3/3
## [1] 0.00000000000000001388
</syntaxhighlight>
 
== Sys.getpid() ==
This can be used to monitor R process memory usage or stop the R process. See [https://stat.ethz.ch/pipermail/r-devel/2016-November/073360.html this post].
 
== How to write R codes ==
* [https://youtu.be/7oyiPBjLAWY Code smells and feels] from R Consortium
** write simple conditions,
** handle class properly,
** return and exit early,
** polymorphism,
** switch(),
** case_when(),
** %||%.
 
== How to debug an R code ==
=== Using assign() in functions ===
For example, insert the following line to your function
<pre>
<pre>
assign(envir=globalenv(), "GlobalVar", localvar)
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
</pre>
</pre>


== Debug lapply()/sapply() ==
* Example 3. Default arguments are evaluated inside the function.
* https://stackoverflow.com/questions/1395622/debugging-lapply-sapply-calls
<pre>
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/recover.html. Use options(error=NULL) to turn it off.
f <- function(x = ls()) {
  a <- 1
  x
}


== Debugging with RStudio ==
# ls() evaluated inside f:
* https://www.rstudio.com/resources/videos/debugging-techniques-in-rstudio/
f()
* https://github.com/ajmcoqui/debuggingRStudio/blob/master/RStudio_Debugging_Cheatsheet.pdf
# [1] "a" "x"
* https://support.rstudio.com/hc/en-us/articles/205612627-Debugging-with-RStudio


== Debug R source code ==
# ls() evaluated in global environment:
=== Build R with debug information ===
f(ls())
* [[R#Build_R_from_its_source|R -> Build R from its source on Windows]]
# [1] "add"    "adders" "f"
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/ (defunct)
</pre>
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/gdb.shtml (defunct)
 
* Build R with debug information (see the discussion [https://stackoverflow.com/a/30001096 here]). Cf [https://github.com/arraytools/r-build-output output messages] from running ./configure and make using the default options.
* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
: <syntaxhighlight lang='bash'>
<pre>
$ ./configure --help
x <- NULL
$ ./configure --enable-R-shlib --with-valgrind-instrumentation=2 \
if (!is.null(x) && x > 0) {
                              --with-system-valgrind-headers \
              CFLAGS='-g -O0 -fPIC' \
              FFLAGS='-g -O0 -fPIC' \
              CXXFLAGS='-g -O0 -fPIC' \
              FCFLAGS='-g -O0 -fPIC'
$ make -j4
$ sudo make install
</syntaxhighlight>
* [https://github.com/arraytools/r-debug My note of debugging cor() function]
* [https://vimeo.com/11937905 Using gdb to debug R packages with native code] (Video) The steps to debug is given below.  
: <syntaxhighlight lang='bash'>
# Make sure to create a file <src/Makevars> with something like: CFLAGS=-ggdb -O0
# Or more generally
# CFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# CXXFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# FFLAGS=-Wall -Wextra -pedantic -O0 -ggdb


$ tree nidemo
}
$ R CMD INSTALL nidemo
</pre>
$ cat bug.R
$ R -f bug.R
$ R -d gdb
(gdb) r
> library(nidemo)
> Ctrl+C
(gdb) b nid_buggy_freq
(gdb) c  # continue
> buggy_freq("nidemo/DESCRIPTION") # stop at breakpoint 1
(gdb) list
(gdb) n # step through
(gdb) # press RETURN a few times until you see the bug
(gdb) d 1 # delete the first break point
(gdb) b Rf_error # R's C entry point for the error function
(gdb) c
> buggy_freq("nidemo/DESCRIPTION")
(gdb) bt 5 # last 5 stack frames
(gdb) frame 2
(gdb) list
(gdb) p freq_data
(gdb) p ans
(gdb) call Rf_PrintValues(ans)
(gdb) call Rf_PrintValues(fname)
(gdb) q
# Edit buggy.c


$ R CMD INSTALL nidemo # re-install the package
=== Use of functions as arguments ===
$ R -f bug.R
[https://www.njtierney.com/post/2019/09/29/unexpected-function/ Just Quickly: The unexpected use of functions as arguments]
$ R -d gdb
(gdb) run
> source("bug.R") # error happened
(gdb) bt 5 # show the last 5 frames
(gdb) frame 2
(gdb) list
(gdb) frame 1
(gdb) list
(gdb) p file
(gdb) p fh
(gdb) q
# Edit buggy.c


$ R CMD INSTALL nidemo
=== body() ===
$ R -f bug.R
[https://stackoverflow.com/a/51548945 Remove top axis title base plot]
</syntaxhighlight>
* [http://r-pkgs.had.co.nz/src.html Compiled code] from "R packages" by Hadley Wickham
* [https://www.bioconductor.org/developers/how-to/c-debugging/ Debugging C/C++ code] from Bioconductor (case study)
* Same idea for the Rcpp situation. See [https://stackoverflow.com/questions/21226337/what-are-productive-ways-to-debug-rcpp-compiled-code-loaded-in-r-on-os-x-maveri What are productive ways to debug Rcpp compiled code loaded in R (on OS X Mavericks)?]


=== .Call ===
=== Return functions in R ===
* [https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Calling-_002eCall Writing R Extensions] manual.
* [https://win-vector.com/2015/04/03/how-and-why-to-return-functions-in-r/ How and why to return functions in R]
* [http://adv-r.had.co.nz/C-interface.html R’s C interface] from Advanced R by Hadley Wickham
* See the doc & example from [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/taskCallback taskCallback - Create an R-level task callback manager]. [https://developer.r-project.org/TaskHandlers.pdf Top-level Task Callbacks in R].
* [https://purrple.cat/blog/2017/05/28/turn-r-users-insane-with-evil/ Turn R users insane with evil]


=== Registering native routines ===
=== anonymous function ===
https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Registering-native-routines
In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a '''lambda function is defined without a name''', while a regular function is defined with a name.


Pay attention to the prefix argument '''.fixes''' (eg .fixes = "C_") in '''useDynLib()''' function in the NAMESPACE file.
<ul>
<li>See [[Tidyverse#Anonymous_functions|Tidyverse]] page
<li>But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, '''lapply(list(1,2,3), function(x) { x * x }) '''
<li>you can use lambda functions with many other functions in R that take a function as an argument. Some examples include '''sapply, apply, vapply, mapply, Map, Reduce, Filter''', and '''Find'''. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
<pre>
Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
</pre>
<li>[https://coolbutuseless.github.io/2019/03/13/anonymous-functions-in-r-part-1/ purrr anonymous function]
<li>[https://towardsdatascience.com/the-new-pipe-and-anonymous-function-syntax-in-r-54d98861014c The new pipe and anonymous function syntax in R 4.1.0]
<li>[http://adv-r.had.co.nz/Functional-programming.html#anonymous-functions Functional programming] from Advanced R
<li>[https://www.projectpro.io/recipes/what-are-anonymous-functions-r What are anonymous functions in R].  
<syntaxhighlight lang='rsplus'>
> (function(x) x * x)(3)
[1] 9
> (\(x) x * x)(3)
[1] 9
</syntaxhighlight>
</ul>


=== Example of debugging cor() function ===
== Backtick sign, infix/prefix/postfix operators ==
Note that R's cor() function called a C function cor().
The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in [http://adv-r.had.co.nz/Functions.html Advanced R] and [https://stackoverflow.com/a/36229703 What do backticks do in R?].
<pre>
<pre>
stats::cor
iris %>% `[[`("Species")
....
  .Call(C_cor, x, y, na.method, method == "kendall")
</pre>
</pre>


A step-by-step screenshot of debugging using the GNU debugger '''gdb''' can be found on my Github repository https://github.com/arraytools/r-debug.
'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
<pre>
1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix
</pre>


== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
Use with functions like sapply, e.g. '''sapply(1:5, `+`, 3) '''  .
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264


== Path length in dir.create() (PR#17206) ==
== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)
<ul>
<li>http://adv-r.had.co.nz/Exceptions-Debugging.html </li>
<li>[https://www.r-bloggers.com/2023/11/catch-me-if-you-can-exception-handling-in-r/ Catch Me If You Can: Exception Handling in R] </li>
<li>Temporarily disable warning messages
<pre>
# Method1:
suppressWarnings(expr)


== install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths() ==
# Method 2:
* https://support.rstudio.com/hc/en-us/community/posts/115008369408-Since-update-to-R-3-4-1-R-LIBS-USER-is-empty and http://r.789695.n4.nabble.com/R-LIBS-USER-on-Ubuntu-16-04-td4740935.html. Modify '''/etc/R/Renviron''' (if you have a sudo right) by uncomment out line 43.
<pre>
<pre>
R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
defaultW <- getOption("warn")
options(warn = -1)
[YOUR CODE]
options(warn = defaultW)
</pre>
</pre>
* https://stackoverflow.com/questions/44873972/default-r-personal-library-location-is-null. Modify '''$HOME/.Renviron''' by adding a line
</li>
<li>try() allows execution to continue even after an error has occurred. You can suppress the message with '''try(..., silent = TRUE)'''.
<pre>
<pre>
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"
out <- try({
</pre>
  a <- 1
* http://stat.ethz.ch/R-manual/R-devel/library/base/html/libPaths.html. Play with .libPaths()
  b <- "x"
 
  a + b
On Mac & R 3.4.0 (it's fine)
})
<syntaxhighlight lang='rsplus'>
> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
</syntaxhighlight>


On Linux & R 3.3.1 (ARM)
elements <- list(1:10, c(-1, 10), c(T, F), letters)
<syntaxhighlight lang='rsplus'>
results <- lapply(elements, log)
> Sys.getenv("R_LIBS_USER")
is.error <- function(x) inherits(x, "try-error")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
succeeded <- !sapply(results, is.error)
> .libPaths()
</pre>
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
</li>
[2] "/usr/local/lib/R/library"
<li>tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
</syntaxhighlight>
<pre>
tryCatch(expr, ..., finally)


On Linux & R 3.4.1 (*Problematic*)
show_condition <- function(code) {
<syntaxhighlight lang='rsplus'>
  tryCatch(code,
> Sys.getenv("R_LIBS_USER")
    error = function(c) "error",
[1] ""
    warning = function(c) "warning",
> .libPaths()
    message = function(c) "message"
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
  )
[3] "/usr/lib/R/library"
}
</syntaxhighlight>
show_condition(stop("!"))
 
#> [1] "error"
I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
show_condition(warning("?!"))
<syntaxhighlight lang='rsplus'>
#> [1] "warning"
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
show_condition(message("?"))
> library(devtools)
#> [1] "message"
Error in library(devtools) : there is no package called 'devtools'
show_condition(10)
#> [1] 10
</pre>
Below is another snippet from available.packages() function,
{{Pre}}
z <- tryCatch(download.file(....), error = identity)
if (!inherits(z, "error")) STATEMENTS
</pre>
</li>
<li>The return class from tryCatch() may not be fixed.
<pre>
result <- tryCatch({
  # Code that might generate an error or warning
  log(99)
}, warning = function(w) {
  # Code to handle warnings
  print(paste("Warning:", w))
}, error = function(e) {
  # Code to handle errors
  print(paste("Error:", e))
}, finally = {
  # Code to always run, regardless of whether an error or warning occurred
  print("Finished")
}) 
# character type. But if we remove 'finally', it will be numeric.
</pre>
<li>[https://www.bangyou.me/post/capture-logs/ Capture message, warnings and errors from a R function]
</li>
</ul>
 
=== suppressMessages() ===
suppressMessages(expression)


# Specify lib.loc parameter will not help with the dependency package
== List data type ==
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
=== Create an empty list ===
Error: package or namespace load failed for 'devtools':
<pre>
.onLoad failed in loadNamespace() for 'devtools', details:
out <- vector("list", length=3L) # OR out <- list()
  call: loadNamespace(name)
for(j in 1:3) out[[j]] <- myfun(j)
  error: there is no package called 'withr'


# A solution is to redefine .libPaths
outlist <- as.list(seq(nfolds))
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
</pre>
> library(devtools) # Works
</syntaxhighlight>


A better solution is to specify R_LIBS_USER in '''~/.Renviron''' file or '''~/.bash_profile'''; see [http://stat.ethz.ch/R-manual/R-patched/library/base/html/Startup.html ?Startup].
=== Using $ in R on a List ===
[https://r-lang.com/dollar-sign-in-r-with-example/ How to Use Dollar sign in R]


== Using external data from within another package ==
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5
</pre>


== How to run R scripts from the command line ==
=== Descend recursively through lists ===
https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line
<nowiki>x[[c(5,3)]] </nowiki> is the same as <nowiki>x[[5]][[3]]</nowiki>. See [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Extract ?Extract].


== How to exit a sourced R script ==
=== Avoid if-else or switch ===
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script How to exit a sourced R script]
?plot.stepfun.
* [http://r.789695.n4.nabble.com/Problem-using-the-source-function-within-R-functions-td907180.html Problem using the source-function within R-functions] ''' ''The best way to handle the generic sort of problem you are describing is to take those source'd files, and rewrite their content as functions to be called from your other functions.'' '''
<pre>
y0 <- c(1,2,4,3)
sfun0  <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)
 
tt <- seq(0, 3, by = 0.1)
op <- par(mfrow = c(2,2))
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
    main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")
 
for(i in 1:3)
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)
 
par(op)
</pre>
[[:File:StepfunExample.svg]]


== Decimal point & decimal comma ==
== Open a new Window device ==
Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark
X11() or dev.new()


== setting seed locally (not globally) in R ==
== par() ==
https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r
?par


== R's internal C API ==
=== text size (cex) and font size on main, lab & axis ===
https://github.com/hadley/r-internals
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]


== cleancall package for C resource cleanup ==
Examples (default is 1 for each of them):
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]
* cex.main=0.9
* cex.sub
* cex.lab=0.8, font.lab=2 (x/y axis labels)
* cex.axis=0.8, font.axis=2 (axis/tick text/labels)
* col.axis="grey50"


== Random numbers: multivariate normal ==
An quick example to increase font size ('''cex.lab''', '''cex.axis''', '''cex.main''') and line width ('''lwd''') in a line plot and '''cex''' & '''lwd''' in the legend.
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
<pre>
plot(x=x$mids, y=x$density, type="l",
    xlab="p-value", ylab="Density", lwd=2,
    cex.lab=1.5, cex.axis=1.5,
    cex.main=1.5, main = "")
lines(y$mids, y$density, lty=2, pwd=2)
lines(z$mids, z$density, lty=3, pwd=2)
legend('topright',legend = c('Method A','Method B','Method C'),
      lty=c(2,1,3), lwd=c(2,2,2), cex = 1.5, xjust = 0.5, yjust = 0.5)
</pre>


The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See  
ggplot2 case (default font size is [https://ggplot2.tidyverse.org/articles/faq-customising.html 11 points]):
* plot.title
* plot.subtitle
* axis.title.x, axis.title.y: (x/y axis labels)
* axis.text.x & axis.text.y: (axis/tick text/labels)
<pre>
ggplot(df, aes(x, y)) +
  geom_point() +
  labs(title = "Title", subtitle = "Subtitle", x = "X-axis", y = "Y-axis") +
  theme(plot.title = element_text(size = 20),
        plot.subtitle = element_text(size = 15),
        axis.title.x = element_text(size = 15),
        axis.title.y = element_text(size = 15),
        axis.text.x = element_text(size = 10),
        axis.text.y = element_text(size = 10))
</pre>
 
=== Default font ===
* [https://stat.ethz.ch/R-manual/R-devel/library/grDevices/html/png.html ?png].  The default font family is '''Arial''' on Windows and '''Helvetica''' otherwise.
* ''sans''. See [https://www.r-bloggers.com/2015/08/changing-the-font-of-r-base-graphic-plots/ Changing the font of R base graphic plots]
* [http://www.cookbook-r.com/Graphs/Fonts/ Fonts] from ''Cookbook for R''. It seems ggplot2 also uses '''sans''' as the default font.
* [https://www.r-bloggers.com/2021/07/using-different-fonts-with-ggplot2/ Using different fonts with ggplot2]
* [https://r-coder.com/plot-r/#Font_family R plot font family]
* [https://r-coder.com/custom-fonts-r/ Add custom fonts in R]
 
=== layout ===
* [https://blog.rsquaredacademy.com/data-visualization-with-r-combining-plots/ Data Visualization with R - Combining Plots]
* http://datascienceplus.com/adding-text-to-r-plot/
 
=== reset the settings ===
{{Pre}}
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)
....
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
</pre>
 
=== mtext (margin text) vs title ===
* https://datascienceplus.com/adding-text-to-r-plot/
* https://datascienceplus.com/mastering-r-plot-part-2-axis/
 
=== mgp (axis tick label locations or axis title) ===
# The margin line (in ‘mex’ units) for the axis title, axis labels and axis line.  Note that ‘mgp[1]’ affects the axis ‘title’ whereas ‘mgp[2:3]’ affect tick mark labels.  The default is ‘c(3, 1, 0)’. If we like to make the axis labels closer to an axis, we can use mgp=c(1.5, .5, 0) for example.
#* the default is c(3,1,0) which specify the margin line for the '''axis title''', '''axis labels''' and '''axis line'''.
#* the axis title is drawn in the fourth line of the margin starting from the plot region, the axis labels are drawn in the second line and the axis line itself is the first line.
# [https://www.r-bloggers.com/2010/06/setting-graph-margins-in-r-using-the-par-function-and-lots-of-cow-milk/ Setting graph margins in R using the par() function and lots of cow milk]
# [https://statisticsglobe.com/move-axis-label-closer-to-plot-in-base-r Move Axis Label Closer to Plot in Base R (2 Examples)]
# http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the '''labels/axis title''' (i.e. xlab and ylab in plot), the second the '''tick-mark labels''', and third the '''tick marks'''. The default is c(3, 1, 0).
 
=== move axis title closer to axis ===
* [https://r-charts.com/base-r/title/ Setting a title and a subtitle]. Default is around 1.7(?). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/title ?title].
* [https://stackoverflow.com/a/30265996 move axis label closer to axis] '''title(, line)'''. This is useful when we use '''xaxt='n' ''' to hide the ticks and labels.
<pre>
title(ylab="Within-cluster variance", line=0,
      cex.lab=1.2, family="Calibri Light")
</pre>
 
=== pch and point shapes ===
[[:File:R pch.png]]
 
See [https://www.statmethods.net/advgraphs/parameters.html here].
 
* Full circle: pch=16
* Display all possibilities: ggpubr::show_point_shapes()
 
=== lty (line type) ===
[[:File:R lty.png]]
 
[https://finnstats.com/index.php/2021/06/11/line-types-in-r-lty-for-r-baseplot-and-ggplot/ Line types in R: Ultimate Guide For R Baseplot and ggplot]
 
See [http://www.sthda.com/english/wiki/line-types-in-r-lty here].
 
ggpubr::show_line_types()
 
=== las (label style) ===
0: The default, parallel to the axis
 
1: Always horizontal
 
2: Perpendicular to the axis
 
3: Always vertical
 
=== oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots ===
<ul>
<li>The following trick is useful when we want to draw multiple plots with a common title.
{{Pre}}
par(mfrow=c(1,2),oma = c(0, 0, 2, 0))  # oma=c(0, 0, 0, 0) by default
plot(1:10,  main="Plot 1")
plot(1:100,  main="Plot 2")
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
</pre>
<li>[[PCA#Visualization|PCA plot]] example (the plot in the middle)
<li>For scatterplot3d() function, '''oma''' is not useful and I need to use '''xpd'''.
<li>[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
<li>[https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par] about '''xpd''' option
* If FALSE (default), all plotting is clipped to the plot region,
* If TRUE, all plotting is clipped to the figure region,
* If NA, all plotting is clipped to the device region.
<li>3 types of regions. See [https://www.benjaminbell.co.uk/2018/02/creating-multi-panel-plots-and-figures.html Creating multi-panel plots and figures using layout()] & [https://www.seehuhn.de/blog/122 publication-quality figures with R, part 2]
* plot region,
* figure region,
* device region.
<li>[https://www.benjaminbell.co.uk/2018/02/creating-multi-panel-plots-and-figures.html Creating multi-panel plots and figures using layout()] includes several tricks including creating a picture-in-picture plot.
</ul>
 
=== no.readonly ===
[https://www.zhihu.com/question/54116933 R语言里par(no.readonly=TURE)括号里面这个参数什么意思?], [https://www.jianshu.com/p/a716db5d30ef R-par()]
 
== Non-standard fonts in postscript and pdf graphics ==
https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41
 
 
== NULL, NA, NaN, Inf ==
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
 
== save()/load() vs saveRDS()/readRDS() vs dput()/dget() vs dump()/source() ==
# saveRDS() can only save one R object while save() does not have this constraint.
# saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See [http://www.fromthebottomoftheheap.net/2012/04/01/saving-and-loading-r-objects/ this post].
<pre>
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)
</pre>
 
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/dput dput]: Writes an ASCII text representation of an R object. The object name is not written (unlike '''dump''').
{{Pre}}
$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato",
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos",
"ast", "trig", "platelet", "protime", "stage")
 
> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5,
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L,
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")
</pre>
 
=== User 'verbose = TRUE' in load() ===
When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.
 
=== What are RDS files anyways ===
[https://www.statworx.com/de/blog/archive-existing-rds-files/ Archive Existing RDS Files]
 
== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
* ==: exact match
* '''all.equal''': compare R objects x and y testing ‘near equality’
* identical: The safe and reliable way to test two objects for being exactly equal.
{{Pre}}
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE
</pre>
 
Be careful about using "==" to return an index of matches in the case of data with missing values.
<pre>
R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1
</pre>
 
See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
 
I found a case when I compare two objects where 1 is generated in ''Linux'' and the other is generated in ''macOS'' that identical() gives FALSE but '''all.equal()''' returns TRUE. The difference has a magnitude only e-17.
 
=== waldo ===
* https://waldo.r-lib.org/ or [https://cloud.r-project.org/web/packages/waldo/index.html CRAN]. Find and concisely describe the difference between a pair of R objects.
* [https://predictivehacks.com/how-to-compare-objects-in-r/ How To Compare Objects In R]
 
=== diffobj: Compare/Diff R Objects ===
https://cran.r-project.org/web/packages/diffobj/index.html
 
== testthat ==
* https://github.com/r-lib/testthat
* [http://www.win-vector.com/blog/2019/03/unit-tests-in-r/ Unit Tests in R]
* [https://davidlindelof.com/machine-learning-in-r-start-with-an-end-to-end-test/ Start with an End-to-End Test]
* [https://www.r-bloggers.com/2023/12/a-beautiful-mind-writing-testable-r-code/ A Beautiful Mind: Writing Testable R Code]
 
== tinytest ==
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
 
[https://cran.r-project.org/web/packages/ttdo/index.html ttdo] adds support of the 'diffobj' package for 'diff'-style comparison of R objects.
 
== Numerical Pitfall ==
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
{{Pre}}
.1 - .3/3
## [1] 0.00000000000000001388
</pre>
 
== Sys.getpid() ==
This can be used to monitor R process memory usage or stop the R process. See [https://stat.ethz.ch/pipermail/r-devel/2016-November/073360.html this post].
 
== Sys.getenv() & make the script more portable ==
Replace all the secrets from the script and replace them with '''Sys.getenv("secretname")'''. You can save the secrets in an '''.Renviron''' file next to the script in the same project.
<pre>
$ for v in 1 2; do MY=$v Rscript -e "Sys.getenv('MY')"; done
[1] "1"
[1] "2"
$ echo $MY
2
</pre>
 
== How to write R codes ==
* [https://youtu.be/7oyiPBjLAWY Code smells and feels] from R Consortium
** write simple conditions,
** handle class properly,
** return and exit early,
** polymorphism,
** switch() [e.g., switch(var, value1=out1, value2=out2, value3=out3). Several examples in [https://github.com/cran/glmnet/blob/master/R/assess.glmnet.R#L103 glmnet] ]
** case_when(),
** %||%.
* [https://appsilon.com/write-clean-r-code/ 5 Tips for Writing Clean R Code] – Leave Your Code Reviewer Commentless
** Comments
** Strings
** Loops
** Code Sharing
**Good Programming Practices
 
== How to debug an R code ==
[[Debug#R|Debug R]]
 
== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264
 
== Path length in dir.create() (PR#17206) ==
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)
 
== install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths() ==
* https://support.rstudio.com/hc/en-us/community/posts/115008369408-Since-update-to-R-3-4-1-R-LIBS-USER-is-empty and http://r.789695.n4.nabble.com/R-LIBS-USER-on-Ubuntu-16-04-td4740935.html. Modify '''/etc/R/Renviron''' (if you have a sudo right) by uncomment out line 43.
<pre>
R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
</pre>
* https://stackoverflow.com/questions/44873972/default-r-personal-library-location-is-null. Modify '''$HOME/.Renviron''' by adding a line
<pre>
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"
</pre>
* http://stat.ethz.ch/R-manual/R-devel/library/base/html/libPaths.html. Play with .libPaths()
 
On Mac & R 3.4.0 (it's fine)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
</pre>
 
On Linux & R 3.3.1 (ARM)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"
</pre>
 
On Linux & R 3.4.1 (*Problematic*)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] ""
> .libPaths()
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"
</pre>
 
I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
{{Pre}}
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
Error in library(devtools) : there is no package called 'devtools'
 
# Specify lib.loc parameter will not help with the dependency package
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
Error: package or namespace load failed for 'devtools':
.onLoad failed in loadNamespace() for 'devtools', details:
  call: loadNamespace(name)
  error: there is no package called 'withr'
 
# A solution is to redefine .libPaths
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works
</pre>
 
A better solution is to specify R_LIBS_USER in '''~/.Renviron''' file or '''~/.bash_profile'''; see [http://stat.ethz.ch/R-manual/R-patched/library/base/html/Startup.html ?Startup].
 
== Using external data from within another package ==
https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/
 
== How to run R scripts from the command line ==
https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line
 
== How to exit a sourced R script ==
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script How to exit a sourced R script]
* [http://r.789695.n4.nabble.com/Problem-using-the-source-function-within-R-functions-td907180.html Problem using the source-function within R-functions] ''' ''The best way to handle the generic sort of problem you are describing is to take those source'd files, and rewrite their content as functions to be called from your other functions.'' '''
* ‘source()’ and ‘example()’ have a new optional argument ‘catch.aborts’ which allows continued evaluation of the R code after an error. [https://developer.r-project.org/blosxom.cgi/R-devel/2023/10/11 4-devel] 2023/10/11.
 
== Decimal point & decimal comma ==
Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark
 
== setting seed locally (not globally) in R ==
https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r
 
== R's internal C API ==
https://github.com/hadley/r-internals
 
== cleancall package for C resource cleanup ==
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]
 
== Random number generator ==
* https://cran.r-project.org/doc/manuals/R-exts.html#Random-numbers
* [https://stackoverflow.com/a/14555220 C code from R with .C(): random value is the same every time]
* [https://arxiv.org/pdf/2003.08009v2.pdf Random number generators produce collisions: Why, how many and more] Marius Hofert 2020 and the published paper in [https://www.tandfonline.com/doi/full/10.1080/00031305.2020.1782261 American Statistician] (including R code).
* R package examples. [https://github.com/cran/party/blob/5ddbd382f01fef2ab993401b43d1fc78d0b061fb/src/RandomForest.c party] package.
 
{{Pre}}
#include <R.h>
 
void myunif(){
  GetRNGstate();
  double u = unif_rand();
  PutRNGstate();
  Rprintf("%f\n",u);
}
</pre>
 
<pre>
$ R CMD SHLIB r_rand.c
$ R
R> dyn.load("r_rand.so")
R> set.seed(1)
R> .C("myunif")
0.265509
list()
R> .C("myunif")
0.372124
list()
R> set.seed(1)
R> .C("myunif")
0.265509
list()
</pre>
 
=== Test For Randomness ===
* [https://predictivehacks.com/how-to-test-for-randomness/ How To Test For Randomness]
* [https://www.r-bloggers.com/2021/08/test-for-randomness-in-r-how-to-check-dataset-randomness/ Test For Randomness in R-How to check Dataset Randomness]
 
== Different results in Mac and Linux ==
=== Random numbers: multivariate normal ===
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
 
The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See  
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
* [https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
<ul>
<syntaxhighlight lang='rsplus'>
<li>[https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
set.seed(1234)
 
junk <- biospear::simdata(n=500, p=500, q.main = 10, q.inter = 10,
See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754 this example]. A little more investigation shows the eigen values differ a little bit on macOS and Linux. See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754#file-mvtnorm_debug-r here].
                          prob.tt = .5, m0=1, alpha.tt= -.5,
</li>
                          beta.main= -.5, beta.inter= -.5, b.corr = .7, b.corr.by=25,
</ul>
                          wei.shape = 1, recr=3, fu=2, timefactor=1)
 
## Method 1: MASS::mvrnorm()
== rle() running length encoding ==
## This is simdata() has used. It gives different numbers on different OS.
* https://en.wikipedia.org/wiki/Run-length_encoding
##
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
library(MASS)
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
set.seed(1234)
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
m0 <-1
* https://rosettacode.org/wiki/Run-length_encoding
n <- 500
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
prob.tt <- .5
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package
p <- 500
 
b.corr.by <- 25
== citation() ==
b.corr <- .7
{{Pre}}
data <- data.frame(treat = rbinom(n, 1, prob.tt) - 0.5)
citation()
n.blocks <- p%/%b.corr.by
citation("MASS")
covMat <- diag(n.blocks) %x%
toBibtex(citation())
  matrix(b.corr^abs(matrix(1:b.corr.by, b.corr.by, b.corr.by, byrow = TRUE) -  
</pre>
                    matrix(1:b.corr.by, b.corr.by, b.corr.by)), b.corr.by, b.corr.by)
 
diag(covMat) <- 1
== R not responding request to interrupt stop process ==
data <- cbind(data, mvrnorm(n, rep(0, p), Sigma = covMat))
[https://stackoverflow.com/a/43172530 R not responding request to interrupt stop process]. ''R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts.'' It seems to match with the case I'm running (''dist()'' function).
range(data)
 
# Mac: -4.963827  4.133723
== Monitor memory usage ==
# Linux/Windows: -4.327635  4.408097
* x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
packageVersion("MASS")
* Windows: memory.size(max=TRUE)
# Mac: [1] ‘7.3.49’
* Linux  
# Linux: [1] ‘7.3.49’
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.  
# Windows: [1] ‘7.3.47’
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
 
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
R.version$version.string
* '''mem_used()''' function from [https://cran.r-project.org/web/packages/pryr/index.html pryr] package. It is not correct or useful if I use it to check the value compared to the memory returned by '''jobload''' in biowulf. So I cannot use it to see the memory used in running mclapply().
# Mac: [1] "R version 3.4.3 (2017-11-30)"
* [https://cran.r-project.org/web/packages/peakRAM/index.html peakRAM]: Monitor the Total and Peak RAM Used by an Expression or Function
# Linux: [1] "R version 3.4.4 (2018-03-15)"
* [https://www.zxzyl.com/archives/1456/ Error: protect () : protection stack overflow] and [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Memory ?Memory]
# Windows: [1] "R version 3.4.3 (2017-11-30)"
 
 
References:
## Method 2: mvtnorm::rmvnorm()
* [https://unix.stackexchange.com/questions/554/how-to-monitor-cpu-memory-usage-of-a-single-process How to monitor CPU/memory usage of a single process?]. ''htop -p $PID'' is recommended. It only shows the percentage of memory usage.
library(mvtnorm)
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
set.seed(1234)
* [https://serverfault.com/a/264856 How can I see the memory usage of a Linux process?] ''pmap $PID | tail -n 1'' is recommended. It shows the memory usage in absolute value (eg 1722376K).
sigma <- matrix(c(4,2,2,3), ncol=2)
* [https://stackoverflow.com/a/6457769 How to check the amount of RAM in R] '''memfree <- as.numeric(system("awk '/MemFree/ {print $2}' /proc/meminfo", intern=TRUE)); memfree '''
x <- rmvnorm(n=n, rep(0, p), sigma=covMat)
 
range(x)
== Monitor Data ==
# Mac: [1] -4.482566  4.459236
[https://www.jstatsoft.org/article/view/v098i01?s=09 Monitoring Data in R with the lumberjack Package]
# Linux: [1] -4.482566  4.459236
 
 
== Pushover ==
## Method 3: mvnfast::rmvn()
[https://rud.is/b/2020/01/29/monitoring-website-ssl-tls-certificate-expiration-times-with-r-openssl-pushoverr-and-dt/ Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}]
set.seed(1234)
 
x <- mvnfast::rmvn(n, rep(0, p), covMat)
[https://cran.r-project.org/web/packages/pushoverr/ pushoverr]
range(x)
 
# Mac: [1] -4.323585  4.355666
= Resource =
# Linux: [1] -4.323585  4.355666
== Books ==
 
* [https://forwards.github.io/rdevguide/ R Development Guide] R Contribution Working Group
library(microbenchmark)
* [https://rviews.rstudio.com/2021/11/04/bookdown-org/ An R Community Public Library] 2011-11-04
library(MASS)
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
library(mvtnorm)
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
library(mvnfast)
* [http://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
microbenchmark(v1 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "eigen"),
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
              v2 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "svd"),
* [http://www.amazon.com/Advanced-Chapman-Hall-CRC-Series/dp/1466586966/ref=pd_sim_b_6?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB Advanced R] by Hadley Wickham 2014
              v3 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "chol"),
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
              v4 <- rmvn(n, rep(0, p), covMat),
* [https://b-rodrigues.github.io/fput/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
              v5 <- mvrnorm(n, rep(0, p), Sigma = covMat))
* [http://www.amazon.com/Cookbook-OReilly-Cookbooks-Paul-Teetor/dp/0596809158/ref=pd_sim_b_3?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB R Cookbook] by Paul Teetor
Unit: milliseconds
* [http://www.amazon.com/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
expr      min        lq
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
v1 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "eigen") 296.55374 300.81089
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
v2 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "svd") 461.81867 466.98806
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
v3 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "chol") 118.33759 120.01829
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
v4 <- rmvn(n, rep(0, p), covMat)  66.64675  69.89383
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
v5 <- mvrnorm(n, rep(0, p), Sigma = covMat) 291.19826 294.88038
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
mean    median        uq      max neval  cld
* Data Manipulation with R by Phil Spector
306.72485 301.99339 304.46662 335.6137  100    d
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
478.58536 470.44085 493.89041 571.7990  100    e
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
125.85427 121.26185 122.21361 151.1658  100  b 
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
71.67996  70.52985  70.92923 100.2622  100 a  
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
301.88144 296.76028 299.50839 346.7049  100  c 
<ul>
</syntaxhighlight>
<li>[https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
A little more investigation shows the eigen values differ a little bit on macOS and Linux.
{{Pre}}
<syntaxhighlight lang='rsplus'>
# R 3.4.1
set.seed(1234); x <- mvrnorm(n, rep(0, p), Sigma = covMat)
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
debug(mvrnorm)
setwd("/tmp/efficientR/")
# eS --- macOS
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# eS2 -- Linux
# generated pdf file is located _book/_main.pdf
Browse[2]> range(abs(eS$values - eS2$values))
 
# [1] 0.000000e+00 1.776357e-15
bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
Browse[2]> var(as.vector(eS$vectors))
# generated epub file is located _book/_main.epub.
[1] 0.002000006
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
Browse[2]> var(as.vector(eS2$vectors))
# but it is OK to run in an R terminal
[1] 0.001999987
</pre>
Browse[2]> all.equal(eS$values, eS2$values)
</li>
[1] TRUE
</ul>
Browse[2]> which(eS$values != eS2$values)
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro
  [1]   6  7  8  9  10  11 12  13  14  20  22  23  24  25  26  27  28  29
* [https://rstats.wtf/ What They Forgot to Teach You About R] Jennifer Bryan & Jim Hester
  ...
* [http://knosof.co.uk/ESEUR/ Evidence-based Software Engineering] by Derek M. Jones
[451] 494 495 496 497 499 500
* [https://www.bigbookofr.com/index.html Big Book of R]
Browse[2]> range(abs(eS$vectors - eS2$vectors))
* [https://epirhandbook.com/?s=09 R for applied epidemiology and public health]
[1] 0.0000000 0.5636919
* [http://bendixcarstensen.com/EwR/ Epidemiology with R] and the [https://cran.r-project.org/web/packages/Epi/ Epi] package. [https://rdrr.io/cran/Epi/man/ci.lin.html ci.lin()] function to return the CI from glm() fit.
</syntaxhighlight>
* [https://education.rstudio.com/learn/ RStudio &rarr; Finding Your Way To R]. Beginners/Intermediates/Experts
 
* [https://deepr.gagolewski.com/index.html Deep R Programming]
== rle() running length encoding ==
 
* https://en.wikipedia.org/wiki/Run-length_encoding
== Videos ==
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
[https://www.infoworld.com/article/3411819/do-more-with-r-video-tutorials.html “Do More with R” video tutorials]. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
 
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
=== Webinar ===
* https://rosettacode.org/wiki/Run-length_encoding
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
 
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package
== useR! ==
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]
* [http://www.user2019.fr/ UseR! 2019], [https://github.com/sowla/useR2019-materials tutorial], [https://www.mango-solutions.com/blog/user2019-roundup-workflow-reproducibility-and-friends Better workflow]
* [https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/playlists UseR! 2020 & 2021]
* [https://rviews.rstudio.com/2021/09/09/a-guide-to-binge-watching-r-medicine/ A Guide to Binge Watching R / Medicine 2021]
* [https://t.co/QBZwNoPJsC UseR! 2022]
 
== R consortium ==
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
 
== Blogs, Tips, Socials, Communities ==
* Google: revolutionanalytics In case you missed it
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]
* The R Blog <s>https://developer.r-project.org/Blog/public/</s> https://blog.r-project.org/
* [https://www.dataquest.io/blog/top-tips-for-learning-r-from-africa-rs-shelmith-kariuki/ Top Tips for Learning R from Africa R’s Shelmith Kariuki]
* [https://smach.github.io/R4JournalismBook/HowDoI.html How Do I? …(do that in R)] by Sharon Machlis
* [https://www.t4rstats.com/ Twitter for R programmers]
 
== Bug Tracking System ==
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
 
Use '''sessionInfo()'''.


== citation() ==
== License ==
<syntaxhighlight lang='rsplus'>
[http://www.win-vector.com/blog/2019/07/some-notes-on-gnu-licenses-in-r-packages/ Some Notes on GNU Licenses in R Packages]
citation()
citation("MASS")
toBibtex(citation())
</syntaxhighlight>
 
== Monitor memory usage ==
* Windows: memory.size(max=TRUE)
* Linux
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''


References:
[https://moderndata.plot.ly/why-dash-uses-the-mit-license/ Why Dash uses the mit license (and not a copyleft gpl license)]
* [https://unix.stackexchange.com/questions/554/how-to-monitor-cpu-memory-usage-of-a-single-process How to monitor CPU/memory usage of a single process?]. ''htop -p $PID'' is recommended. It only shows the percentage of memory usage.
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
* [https://serverfault.com/a/264856 How can I see the memory usage of a Linux process?] ''pmap $PID | tail -n 1'' is recommended. It shows the memory usage in absolute value (eg 1722376K).


= Resource =
== Interview questions ==
== Books ==
* Does R store matrices in column-major order or row-major order?
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
** Matrices are stored in column-major order, which means that elements are arranged and accessed by columns. This is in contrast to languages like Python, where matrices (or arrays) are typically stored in row-major order.
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
* [http://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
* [http://www.amazon.com/Advanced-Chapman-Hall-CRC-Series/dp/1466586966/ref=pd_sim_b_6?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB Advanced R] by Hadley Wickham 2014
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
* [http://www.amazon.com/Cookbook-OReilly-Cookbooks-Paul-Teetor/dp/0596809158/ref=pd_sim_b_3?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB R Cookbook] by Paul Teetor
* [http://www.amazon.com/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
* [https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
: <syntaxhighlight lang='rsplus'>
# R 3.4.1
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
setwd("/tmp/efficientR/")
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# generated pdf file is located _book/_main.pdf


bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
* Explain the difference between == and === in R. Provide an example to illustrate their use.
# generated epub file is located _book/_main.epub.
** The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.  
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
# but it is OK to run in an R terminal
</syntaxhighlight>
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro


== Webinar ==
* What is the purpose of the apply() function in R? How does it differ from the for loop?
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
** The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.


== useR! ==
* Describe the concept of factors in R. How are they used in data manipulation and analysis?
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
** Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]


= R consortium =
* What is the significance
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
of the attach() and detach() functions in R? When should they be used?
** A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.


== Blogs, Tips, Socials, Communities ==
* Explain the concept of vectorization in R. How does it impact the performance of R code?
* Google: revolutionanalytics In case you missed it
** Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]


== Bug Tracking System ==
* Describe the difference between data.frame and matrix in R. When would you use one over the other?
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
** A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
** A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
** You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.


Use '''sessionInfo()'''.
* What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
** The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
** It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
** To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.

Revision as of 13:56, 22 April 2024

Install and upgrade R

Here

New release

Online Editor

We can run R on web browsers without installing it on local machines (similar to [/ideone.com Ideone.com] for C++. It does not require an account either (cf RStudio).

rdrr.io

It can produce graphics too. The package I am testing (cobs) is available too.

rstudio.cloud

RDocumentation

The interactive engine is based on DataCamp Light

For example, tbl_df function from dplyr package.

The website DataCamp allows to run library() on the Script window. After that, we can use the packages on R Console.

Here is a list of (common) R packages that users can use on the web.

The packages on RDocumentation may be outdated. For example, the current stringr on CRAN is v1.2.0 (2/18/2017) but RDocumentation has v1.1.0 (8/19/2016).

Web Applications

R web applications

Creating local repository for CRAN and Bioconductor

R repository

Parallel Computing

See R parallel.

Cloud Computing

Install R on Amazon EC2

http://randyzwitch.com/r-amazon-ec2/

Bioconductor on Amazon EC2

http://www.bioconductor.org/help/bioconductor-cloud-ami/

Big Data Analysis

bigmemory, biganalytics, bigtabulate

ff, ffbase

biglm

data.table

See data.table.

disk.frame

Split-apply-combine for Maximum Likelihood Estimation of a linear model

Apache arrow

Reproducible Research

Reproducible Environments

https://rviews.rstudio.com/2019/04/22/reproducible-environments/

checkpoint package

Some lessons in R coding

  1. don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use Rcpp package and R's random number generator instead.
  2. don't use list.files() directly. The result is platform dependent even different Linux OS. An extra sorting helps!

Useful R packages

Rcpp

http://cran.r-project.org/web/packages/Rcpp/index.html. See more here.

RInside : embed R in C++ code

Ubuntu

With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on Youtube. I have tested this qtdensity example successfully using Qt 4.8.5.

  1. Follow the instruction cairoDevice to install required libraries for cairoDevice package and then cairoDevice itself.
  2. Install Qt. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
  3. Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
  4. Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the interactive program qtdensity appears on your desktop.

File:qtdensity.png

With RInside + Wt web toolkit installed, we can also create a web application. To demonstrate the example in examples/wt directory, we can do

cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
make
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080

Then we can go to the browser's address bar and type http://localhost:8080 to see how it works (a screenshot is in here).

Windows 7

To make RInside works on Windows OS, try the following

  1. Make sure R is installed under C:\ instead of C:\Program Files if we don't want to get an error like g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory.
  2. Install RTools
  3. Instal RInside package from source (the binary version will give an error )
  4. Create a DOS batch file containing necessary paths in PATH environment variable
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
cmd

In the Windows command prompt, run

cd C:\R\R-3.0.1\library\RInside\examples\standard
make -f Makefile.win

Now we can test by running any of executable files that make generates. For example, rinside_sample0.

rinside_sample0

As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See some discussions in

So the Qt and Wt web tool applications on Windows may or may not be possible.

GUI

Qt and R

tkrplot

On Ubuntu, we need to install tk packages, such as by

sudo apt-get install tk-dev

reticulate - Interface to 'Python'

Python -> reticulate

Hadoop (eg ~100 terabytes)

See also HighPerformanceComputing

RHadoop

Snowdoop: an alternative to MapReduce algorithm

XML

On Ubuntu, we need to install libxml2-dev before we can install XML package.

sudo apt-get update
sudo apt-get install libxml2-dev

On CentOS,

yum -y install libxml2 libxml2-devel

XML

library(XML)

# Read and parse HTML file
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)

# Extract all the paragraphs (HTML tag is p, starting at
# the root of the document). Unlist flattens the list to
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))

# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)

# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')

This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.

> library(RCurl) # getURL()
> library(XML)   # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # 2.5.3a
> plain.text
 [1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"     
 [6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"

> # try picard
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
 [1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
[10] "2.6.0"

This method can be used to monitor new tags/releases from some projects like Cura, BWA, Picard, STAR. But for some projects like sratools the class attribute in the span element ("css-truncate-target") can be different (such as "tag-name").

xmlview

RCurl

On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)

# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev

Scrape google scholar results

https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R

No google ID is required

Seems not work

 Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  : 
  arguments imply differing number of rows: 2, 0 

devtools

devtools package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the remotes package which was suggested by the BiocManager package.

# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev

# Ubuntu 16.04, 18.04
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev

# Ubuntu 20.04
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev

Lazy-load database XXX is corrupt. internal error -3. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.

NB. According to the output of apt-cache show r-cran-devtools, the binary package is very old though apt-cache show r-base and supported packages like survival shows the latest version.

httr

httr imports curl, jsonlite, mime, openssl and R6 packages.

When I tried to install httr package, I got an error and some message:

Configuration failed because openssl was not found. Try installing:
 * deb: libssl-dev (Debian, Ubuntu, etc)
 * rpm: openssl-devel (Fedora, CentOS, RHEL)
 * csw: libssl_dev (Solaris)
 * brew: openssl (Mac OSX)
If openssl is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’

It turns out after I run sudo apt-get install libssl-dev in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!

Real example: see this post. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).

Since httr package was used in many other packages, take a look at how others use it. For example, aRxiv package.

A package to download free Springer books during Covid-19 quarantine, An update to "An adventure in downloading books" (rvest package)

curl

curl is independent of RCurl package.

library(curl)
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)

rOpenSci packages

rOpenSci contains packages that allow access to data repositories through the R statistical programming environment

remotes

Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).

Example:

# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')

DirichletMultinomial

On Ubuntu, we do

sudo apt-get install libgsl0-dev

Create GUI

gWidgets

GenOrd: Generate ordinal and discrete variables with given correlation matrix and marginal distributions

here

json

R web -> json

Map

leaflet

choroplethr

ggplot2

How to make maps with Census data in R

googleVis

See an example from RJSONIO above.

googleAuthR

Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.

gtrendsR - Google Trends

quantmod

Maintaining a database of price files in R. It consists of 3 steps.

  1. Initial data downloading
  2. Update existing data
  3. Create a batch file

caret

Tool for connecting Excel with R

write.table

Output a named vector

vec <- c(a = 1, b = 2, c = 3)
write.csv(vec, file = "my_file.csv", quote = F)
x = read.csv("my_file.csv", row.names = 1)
vec2 <- x[, 1]
names(vec2) <- rownames(x)
all.equal(vec, vec2)

# one liner: row names of a 'matrix' become the names of a vector
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
all.equal(vec, vec3)

Avoid leading empty column to header

write.table writes unwanted leading empty column to header when has rownames

write.table(a, 'a.txt', col.names=NA)

Add blank field AND column names in write.table

  • write.table(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
    • Suppose x is (n x 2)
    • write.table(x, sep="\t") will generate a file with 2 element on the 1st row
    • read.table(file) will return an object with a size (n x 2)
    • read.delim(file) and read.delim2(file) will also be correct
  • Note that write.csv() does not have this issue that write.table() has
    • Suppose x is (n x 2)
    • Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
    • If we use read.csv(file), the object is (n x 3). So we need to use read.csv(file, row.names = 1)
  • adding blank field AND column names in write.table(); write.table writes unwanted leading empty column to header when has rownames
write.table(a, 'a.txt', col.names=NA)
  • readr::write_tsv() does not include row names in the output file

read.delim(, row.names=1) and write.table(, row.names=TRUE)

How to Use read.delim Function in R

Case 1: no row.names

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...

Case 2: with row.names. Note: if we open the text file in Excel, we'll see the 1st row is missing one header at the end. It is actually missing the column name for the 1st column.

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
my_df <- read.delim('my_data.txt')  # it will automatically assign the rownames

Read/Write Excel files package

  • http://www.milanor.net/blog/?p=779
  • flipAPI. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
  • xlsx: depends on Java
  • openxlsx: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
  • readxl: it does not depend on anything although it can only read but not write Excel files.
    • It is part of tidyverse package. The readxl website provides several articles for more examples.
    • readxl webinar.
    • One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a tbl_df, a tbl and a data.frame.
    • For writing to Excel formats, use writexl or openxlsx package.
library(readxl)
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE, 
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf, 
    guess_max = min(1000, n_max), progress = readxl_progress(), 
    .name_repair = "unique")
# Example
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
  • writexl: zero dependency xlsx writer for R
library(writexl)
mylst <- list(sheet1name = df1, sheet2name = df2)
write_xlsx(mylst, "output.xlsx")

For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).

> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                   HK1A1   21652 Hs.445981
2                   HK1A2   22012 Hs.119177
3                   HK1A4   22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1     1495
2                              ADP-ribosylation factor 3   ARF3      377
3                          Uroporphyrinogen III synthase   UROS     7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1   5.000000        5q31.2               <NA>      1
2  12.000000         12q13               <NA>      1
3       <NA> 10q25.2-q26.3               <NA>      1

The hidden worksheets become visible (Not sure what are those first rows mean in the output).

> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00 
[1] "Experiment descriptors" "Filtered log ratio"     "Gene identifiers"      
[4] "Gene annotations"       "CollateInfo"            "GeneSubsets"           
[7] "GeneSubsetsTemp"       

The Chinese character works too.

> read_excel("~/Downloads/testChinese.xlsx", 1)
   中文 B C
1     a b c
2     1 2 3

To read all worksheets we need a convenient function

read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc1) is a tibble.

readr

Compared to base equivalents like read.csv(), readr is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.

1.0.0 released. readr 2.0.0 adds built-in support for reading multiple files at once, fast multi-threaded lazy reading and automatic guessing of delimiters among other changes.

Consider a text file where the table (6100 x 22) has duplicated row names and the (1,1) element is empty. The column names are all unique.

  • read.delim() will treat the first column as rownames but it does not allow duplicated row names. Even we use row.names=NULL, it still does not read correctly. It does give warnings (EOF within quoted string & number of items read is not a multiple of the number of columns). The dim is 5177 x 22.
  • readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
  • data.table::fread(Filename, sep = "\t") will detect the number of column names is less than the number of columns. Added 1 extra default column name for the first column which is guessed to be row names or an index. The dim is 6100 x 22. (Winner!)

The readr::read_csv() function is as fast as data.table::fread() function. For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv(). See 5.3 of Efficient R Programming book.

Note that data.table::fread() can read a selection of the columns.

Speed comparison

The Fastest Way To Read And Write Files In R. data.table >> readr >> base.

ggplot2

See ggplot2

Data Manipulation & Tidyverse

See Tidyverse.

Data Science

See Data science page

microbenchmark & rbenchmark

Plot, image

jpeg

If we want to create the image on this wiki left hand side panel, we can use the jpeg package to read an existing plot and then edit and save it.

We can also use the jpeg package to import and manipulate a jpg image. See Fun with Heatmaps and Plotly.

png and resolution

It seems people use res=300 as a definition of high resolution.

  • Bottom line: fix res=300 and adjust height/width as needed. The default is res=72, height=width=480. If we increase res=300, the text font size will be increased, lines become thicker and the plot looks like a zoom-in.
  • Saving high resolution plot in png.
    png("heatmap.png", width = 8, height = 6, units='in', res = 300) 
    # we can adjust width/height as we like
    # the pixel values will be width=8*300 and height=6*300 which is equivalent to 
    # 8*300 * 6*300/10^6 = 4.32 Megapixels (1M pixels = 10^6 pixels) in camera's term
    # However, if we use png(, width=8*300, height=6*300, units='px'), it will produce
    # a plot with very large figure body and tiny text font size.
    
    # It seems the following command gives the same result as above
    png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
    
  • Chapter 14.5 Outputting to Bitmap (PNG/TIFF) Files by R Graphics Cookbook
    • Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.
  • 10 tips for making your R graphics look their best David Smith
    • In Word you can resize the graphic to an appropriate size, but the high resolution gives you the flexibility to choose a size while not compromising on the quality. I'd recommend at least 1200 pixels on the longest side for standard printers.
  • ?png. The png function has default settings ppi=72, height=480, width=480, units="px".
    • By default no resolution is recorded in the file, except for BMP.
    • BMP vs PNG format. If you need a smaller file size and don’t mind a lossless compression, PNG might be a better choice. If you need to retain as much detail as possible and don’t mind a larger file size, BMP could be the way to go.
      • Compression: BMP files are raw and uncompressed, meaning they’re large files that retain as much detail as possible. On the other hand, PNG files are compressed but still lossless. This means you can reduce or expand PNGs without losing any information.
      • File size: BMPs are larger than PNGs. This is because PNG files automatically compress, and can be compressed again to make the file even smaller.
      • Common uses: BMP contains a maximum amount of details while PNGs are good for small illustrations, sketches, drawings, logos and icons.
      • Quality: No difference
      • Transparency: PNG supports transparency while BMP doesn't
  • Some comparison about the ratio
    • 11/8.5=1.29 (A4 paper)
    • 8/6=1.33 (plot output)
    • 1440/900=1.6 (my display)
  • Setting resolution and aspect ratios in R
  • The difference of res parameter for a simple plot. How to change the resolution of a plot in base R?
  • High Resolution Figures in R.
  • High resolution graphics with R
  • R plot: size and resolution
  • How can I increase the resolution of my plot in R?, devEMF package
  • See Images -> Anti-alias.
  • How to check DPI on PNG
    • The width of a PNG file in terms of inches cannot be determined directly from the file itself, as the file contains pixel dimensions, not physical dimensions. However, you can calculate the width in inches if you know the resolution (DPI, dots per inch) of the image. Remember that converting pixel measurements to physical measurements like inches involves a specific resolution (DPI), and different devices may display the same image at different sizes due to having different resolutions.
  • Cairo case.

PowerPoint

  • For PP presentation, I found it is useful to use svg() to generate a small size figure. Then when we enlarge the plot, the text font size can be enlarged too. According to svg, by default, width = 7, height = 7, pointsize = 12, family = sans.
  • Try the following code. The font size is the same for both plots/files. However, the first plot can be enlarged without losing its quality.
    svg("svg4.svg", width=4, height=4)
    plot(1:10, main="width=4, height=4")
    dev.off()
    
    svg("svg7.svg", width=7, height=7) # default
    plot(1:10, main="width=7, height=7")
    dev.off()
    

magick

https://cran.r-project.org/web/packages/magick/

See an example here I created.

Cairo

See White strips problem in png() or tiff().

geDevices

cairoDevice

PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).

For ubuntu OS, we need to install 2 libraries and 1 R package RGtk2.

sudo apt-get install libgtk2.0-dev libcairo2-dev

On Windows OS, we may got the error: unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' . We need to follow the instruction in here.

dpi requirement for publication

For import into PDF-incapable programs (MS Office)

sketcher: photo to sketch effects

https://htsuda.net/sketcher/

httpgd

igraph

creating directed networks with igraph

Identifying dependencies of R functions and scripts

https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts

library(mvbutils)
foodweb(where = "package:batr")

foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)

foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)

iterators

Iterator is useful over for-loop if the data is already a collection. It can be used to iterate over a vector, data frame, matrix, file

Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.

Colors

  • scales package. This is used in ggplot2 package.
  • colorspace: A Toolbox for Manipulating and Assessing Colors and Palettes. Popular! Many reverse imports/suggests; e.g. ComplexHeatmap. See my ggplot2 page.
    hcl_palettes(plot = TRUE) # a quick overview
    hcl_palettes(palette = "Dark 2", n=5, plot = T)
    q4 <- qualitative_hcl(4, palette = "Dark 3")
    
  • convert hex value to color names
    library(plotrix)
    sapply(rainbow(4), color.id) # color.id is a function
              # it is used to identify closest match to a color
    sapply(palette(), color.id)
    sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
    

Below is an example using the option scale_fill_brewer(palette = "Paired"). See the source code at gist. Note that only set1 and set3 palettes in qualitative scheme can support up to 12 classes.

According to the information from the colorbrew website, qualitative schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.

File:GgplotPalette.svg

colortools

Tools that allow users generate color schemes and palettes

colourpicker

A Colour Picker Tool for Shiny and for Selecting Colours in Plots

eyedroppeR

Select colours from an image in R with {eyedroppeR}

rex

Friendly Regular Expressions

formatR

The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.

See also this discussion on stackoverflow talks about R code reformatting.

library(formatR)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
tidy_source("clipboard") 
# default width is getOption("width") which is 127 in my case.

Some issues

  • Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
cat("abcd",
    # This is my comment
    "defg")

will result in

> tidy_source("clipboard")
Error in base::parse(text = code, srcfile = NULL) : 
  3:1: unexpected string constant
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
3: "defg"
   ^
  • Comments appearing at the end of a line within a long complete statement won't break tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
cat("abcd"
    ,"defg"   # This is my comment
  ,"ghij")

will become

cat("abcd", "defg"  # This is my comment
, "ghij") 

Still bad!!

  • Comments appearing at the end of a line within a long complete statement breaks tidy_source() function. For example,
cat("</p>",
	"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
	ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
	                     "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"), 
	file=ExternalFileName, sep="\n", append=T)

will result in

> tidy_source("clipboard", width.cutoff=70)
Error in base::parse(text = code, srcfile = NULL) : 
  3:129: unexpected SPECIAL
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
  • width.cutoff parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on", 
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName, 
    sep = "\n", append = T)
  • It merges lines though I don't always want to do that. For example
cat("abcd"
    ,"defg"  
  ,"ghij")

will become

cat("abcd", "defg", "ghij") 

styler

https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.

Download papers

biorxivr

Search and Download Papers from the bioRxiv Preprint Server (biology)

aRxiv

Interface to the arXiv API

pdftools

aside: set it aside

An RStudio addin to run long R commands aside your current session.

Teaching

  • smovie: Some Movies to Illustrate Concepts in Statistics

Organize R research project

How to save (and load) datasets in R (.RData vs .Rds file)

How to save (and load) datasets in R: An overview

Naming convention

Efficient Data Management in R

Efficient Data Management in R. .Rprofile, renv package and dplyr package.

Text to speech

Text-to-Speech with the googleLanguageR package

Speech to text

https://github.com/ggerganov/whisper.cpp and an R package audio.whisper

Weather data

logR

https://github.com/jangorecki/logR

Progress bar

https://github.com/r-lib/progress#readme

Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.

cron

beepr: Play A Short Sound

https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".

utils package

https://www.rdocumentation.org/packages/utils/versions/3.6.2

tools package

Different ways of using R

Extending R by John M. Chambers (2016)

10 things R can do that might surprise you

https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/

R call C/C++

Mainly talks about .C() and .Call().

Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.

.Call

Be sure to add the PACKAGE parameter to avoid an error like

cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group, 
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) : 
  "standardize" not resolved from current namespace (grpreg)

NAMESPACE file & useDynLib

(From Writing R Extensions manual) Loading is most often done automatically based on the useDynLib() declaration in the NAMESPACE file, but may be done explicitly via a call to library.dynam(). This has the form

library.dynam("libname", package, lib.loc) 

library.dynam.unload()

gcc

Coping with varying `gcc` versions and capabilities in R packages

Primitive functions

Primitive Functions List

SEXP

Some examples from packages

  • sva package has one C code function

R call Fortran

Embedding R

An very simple example (do not return from shell) from Writing R Extensions manual

The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.

This example can be run by

R_HOME/bin/R CMD R_HOME/bin/exec/R

Note:

  1. R_HOME/bin/exec/R is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
  2. R_HOME/bin/R is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to /usr/local/bin/R. When we run R_HOME/bin/R, it actually runs R_HOME/bin/R CMD R_HOME/bin/exec/R (see line 259 of R_HOME/bin/R as in R 3.0.2) so we know the important role of R_HOME/bin/exec/R.

More examples of embedding can be found in tests/Embedding directory. Read <index.html> for more information about these test examples.

An example from Bioconductor workshop

Example: Create embed.c file. Then build the executable. Note that I don't need to create R_HOME variable.

cd 
tar xzvf 
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest

nano embed.c
# Using a single line will give an error and cannot not show the real problem.
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
# A better way is to run compile and link separately
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed

Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying R_HOME variable and including the directories used in linking R in LD_LIBRARY_PATH. This is based on the inform provided by Writing R Extensions.

export R_HOME=/home/brb/Downloads/R-3.0.2
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
./embed # No need to include R CMD in front.

Question: Create a data frame in C? Answer: Use data.frame() via an eval() call from C. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as here.

Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf

Create a Simple Socket Server in R

This example is coming from this paper.

Create an R function

simpleServer <- function(port=6543)
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}

Then run simpleServer(). Open another terminal and try to communicate with the server

$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.

Welcome to R!
R> summary(iris[, 3:5])
  Petal.Length    Petal.Width          Species  
 Min.   :1.000   Min.   :0.100   setosa    :50  
 1st Qu.:1.600   1st Qu.:0.300   versicolor:50  
 Median :4.350   Median :1.300   virginica :50  
 Mean   :3.758   Mean   :1.199                  
 3rd Qu.:5.100   3rd Qu.:1.800                  
 Max.   :6.900   Max.   :2.500                  

R> quit
Connection closed by foreign host.

Rserve

Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See Example from Bioconductor workshop.

See my Rserve page.

outsider

(Commercial) StatconnDcom

R.NET

rJava

Terminal

# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk

and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.

  • Create the file /etc/ld.so.conf.d/java.conf with the following entries:
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
  • And then run sudo ldconfig

Now go back to R

install.packages("rJava")

Done!

If above does not work, a simple way is by (under Ubuntu) running

sudo apt-get install r-cran-rjava

which will create new package 'default-jre' (under /usr/lib/jvm) and 'default-jre-headless'.

RCaller

RApache

Rscript, arguments and commandArgs()

Passing arguments to an R script from command lines Syntax:

$ Rscript --help
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]

Example:

args = commandArgs(trailingOnly=TRUE)
# test if there is at least one argument: if not, return an error
if (length(args)==0) {
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
} else if (length(args)==1) {
  # default output file
  args[2] = "out.txt"
}
cat("args[1] = ", args[1], "\n")
cat("args[2] = ", args[2], "\n")
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt 
# args[2] =  out.txt

Rscript, #! Shebang and optparse package

littler

Provides hash-bang (#!) capability for R

FAQs:

root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"

-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
                                              # Example: R -q -e "date()"

-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
                                              # It won't show the startup message when it is used in the command line.
                                              # Example: Rscript -e "date()"

We can install littler using two ways.

  • install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r). You need to create a soft link in order to access it globally.
  • sudo apt install littler. This will install 'r' globally; however, the installed version may be old.

After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The web version of examples does not have the TOC.

r was not meant to run interactively like R. See man r.

RInside: Embed R in C++

See RInside

(From RInside documentation) The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.

The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.

To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.

To run any executable program, we need to specify LD_LIBRARY_PATH variable, something like

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib 

The real build process looks like (check <Makefile> for completeness)

g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include   \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0

Hello World example of embedding R in C++.

#include <RInside.h>                    // for the embedded R via RInside

int main(int argc, char *argv[]) {

    RInside R(argc, argv);              // create an embedded R instance 

    R["txt"] = "Hello, world!\n";	// assign a char* (string) to 'txt'

    R.parseEvalQ("cat(txt)");           // eval the init string, ignoring any returns

    exit(0);
}

The above can be compared to the Hello world example in Qt.

#include <QApplication.h>
#include <QPushButton.h>

int main( int argc, char **argv )
{
    QApplication app( argc, argv );

    QPushButton hello( "Hello world!", 0 );
    hello.resize( 100, 30 );

    app.setMainWidget( &hello );
    hello.show();

    return app.exec();
}

RFortran

RFortran is an open source project with the following aim:

To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.

It works only on Windows platform with Microsoft Visual Studio installed:(

Call R from other languages

C

Using R from C/C++

Error: “not resolved from current namespace” error, when calling C routines from R

Solution: add getNativeSymbolInfo() around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace

JRI

http://www.rforge.net/JRI/

ryp2

http://rpy.sourceforge.net/rpy2.html

Create a standalone Rmath library

R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of R-admin manual.

Here is my experience based on R 3.0.2 on Windows OS.

Create a static library <libRmath.a> and a dynamic library <Rmath.dll>

Suppose we have downloaded R source code and build R from its source. See Build_R_from_its_source. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.

cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win

Use Rmath library in our code

set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.

# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# It is OK to save the cpp file under any directory.

# Force to link against the static library <libRmath.a>
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe

# Force to link against dynamic library <Rmath.dll>
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe

Test the executable program. Note that the executable program RmathEx1.exe can be transferred to and run in another computer without R installed. Isn't it cool!

c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689

Below is the cpp program <RmathEx1.cpp>.

//RmathEx1.cpp
#define MATHLIB_STANDALONE 
#include <iostream>
#include "Rmath.h"

using std::cout; using std::cin; using std::endl;

int main()
{
  double x1, x2;
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;

  cout << "Prob(Z <= " << x1 << ") = " << 
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " << 
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}

Calling R.dll directly

See Chapter 8.2.2 of R Extensions. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.

Create HTML report

ReportingTools (Jason Hackney) from Bioconductor. See Genome->ReportingTools.

htmlTable package

The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.

formattable

htmltab package

This package is NOT used to CREATE html report but EXTRACT html table.

ztable package

Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.

Create academic report

reports package in CRAN and in github repository. The youtube video gives an overview of the package.

Create pdf and epub files

# Idea:
#        knitr        pdflatex
#   rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
  • A very simple example <002-minimal.Rnw> from yihui.name works fine on linux.
git clone https://github.com/yihui/knitr-examples.git
  • <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run sudo apt-get install texlive-fonts-recommended to install missing fonts. It works!

To see a real example, check out DESeq2 package (inst/doc subdirectory). In addition to DESeq2, I also need to install DESeq, BiocStyle, airway, vsn, gplots, and pasilla packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.

Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.

# Idea:
#        knitr        pandoc
#   rmd -------> md ----------> pdf

git clone https://github.com/yihui/knitr-examples.git
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!

To create an epub file (not success yet on Windows OS, missing figures on Linux OS)

# Idea:
#        knitr        pandoc
#   rnw -------> tex ----------> markdown or epub

library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")

Convert tex to epub

kable() for tables

Create Tables In LaTeX, HTML, Markdown And ReStructuredText

Create Word report

Using the power of Word

How to go from R to nice tables in Microsoft Word

knitr + pandoc

It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.

# Idea:
#        knitr       pandoc
#   rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files

and then

FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))

Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.

Another way is

library(pander)
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")

Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:

  • A pdf file: pandoc -s report.md -t latex -o report.pdf
  • A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
  • Openoffice: pandoc report.md -o report.odt
  • Word docx: pandoc report.md -o report.docx

We can also create the epub file for reading on Kobo ereader. For example, download this file and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!

knit("example.Rmd")
pandoc("example.md", format="epub")

PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)

> pandoc("Rmd_to_Epub.md", format="epub")
executing pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
Error in (function (input, format, ext, cfg)  : conversion failed
In addition: Warning message:
running command 'pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1

pander

Try pandoc[1] with a minimal reproducible example, you might give a try to my "pander" package [2] too:

library(pander)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
            output = tempfile(), convert = 'docx')

Where the content of the "minimal.brew" file is something you might have got used to with Sweave - although it's using "brew" syntax instead. See the examples of pander [3] for more details. Please note that pandoc should be installed first, which is pretty easy on Windows.

  1. http://johnmacfarlane.net/pandoc/
  2. http://rapporter.github.com/pander/
  3. http://rapporter.github.com/pander/#examples

R2wd

Use R2wd package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.

> library(R2wd)
> wdGet()
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
     installstatconnDCOM()

This will download and install the current version of statconnDCOM

You will need a working Internet connection
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() : 
  argument is of length zero 

The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.

Convert from pdf to word

The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert

rtf

Use rtf package for Rich Text Format (RTF) Output.

xtable

Package xtable will produce html output.

print(xtable(X), type="html")

If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.

officer

  • CRAN. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.
  • The gist includes a comprehensive example that encompasses various elements such as sections, subsections, and tables. It also incorporates a detailed paragraph, along with visual representations created using base R plots and ggplots.
  • Add a line space
    doc <- body_add_par(doc, "")
    
    # Function to add n line spaces
    body_add_par_n <- function (doc, n) {
      for(i in 1:n){
        doc <- body_add_par(doc, "")
      }
      return(doc)
    }
    body_add_par_n(3)
    
  • Figures from the documentation of officeverse.
  • See Data frame to word table?.
  • See Office page for some code.
  • How to read and create Word Documents in R where we can extracting tables from Word Documents.
    x = read_docx("myfile.docx")
    content <- docx_summary(x) # a vector
    grep("nlme", content$text, ignore.case = T, value = T)
    

Powerpoint

PDF manipulation

staplr

R Graphs Gallery

COM client or server

Client

RDCOMClient where excel.link depends on it.

Server

RDCOMServer

Use R under proxy

http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy

RStudio

  • Github
  • Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
  • Preview

rstudio.cloud

https://rstudio.cloud/

Launch RStudio

Multiple versions of R

Create .Rproj file

If you have an existing package that doesn't have an .Rproj file, you can use devtools::use_rstudio("path/to/package") to add it.

With an RStudio project file, you can

  • Restore .RData into workspace at startup
  • Save workspace to .RData on exit (or save.image("Robj.RData") & load("Robj.RData"))
  • Always save history (even if no saving .RData, savehistory(".Rhistory") & loadhistory(".Rhistory"))
  • etc

package search

https://github.com/RhoInc/CRANsearcher

Git

Visual Studio

R and Python support now built in to Visual Studio 2017

List files using regular expression

  • Extension
list.files(pattern = "\\.txt$")

where the dot (.) is a metacharacter. It is used to refer to any character.

  • Start with
list.files(pattern = "^Something")

Using Sys.glob()"' as

> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"       "/home/brb/Downloads/valgrind.txt"

Hidden tool: rsync in Rtools

c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe

sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
total size is 1198416  speedup is 3.71

c:\Rtools\bin>

Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also rsync file permissions on windows.

Install rgdal package (geospatial Data) on ubuntu

Terminal

sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411

R

install.packages("rgdal")

Install sf package

I got the following error even I have installed some libraries.

checking GDAL version >= 2.0.1... no
configure: error: sf is not compatible with GDAL versions below 2.0.1

Then I follow the instruction here

sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable

sudo apt update
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears 

sudo apt install libgdal-dev # works on ubuntu 20.04 too
                             # no need the previous lines

Database

RSQLite

Creating a new database:

library(DBI)

mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
dbDisconnect(mydb)
unlink("my-db.sqlite")

# temporary database
mydb <- dbConnect(RSQLite::SQLite(), "")
dbDisconnect(mydb)

Loading data:

mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)

dbListTables(mydb)

dbListFields(con, "mtcars")

dbReadTable(con, "mtcars")

Queries:

dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))

res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
dbFetch(res)

Batched queries:

dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}

dbClearResult(rs)

Multiple parameterised queries:

rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)

Statements:

dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
#> [1] 0
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)

sqldf

Manipulate R data frames using SQL. Depends on RSQLite. A use of gsub, reshape2 and sqldf with healthcare data

RPostgreSQL

RMySQL

MongoDB

odbc

RODBC

DBI

dbplyr

Create a new SQLite database:

surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")

my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)

copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db

Connect to a database:

download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")

library(dbplyr)
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")

Querying the database with the SQL syntax:

tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))

Querying the database with the dplyr syntax:

surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)

show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database

Simple database queries:

surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)

Laziness (instruct R to stop being lazy):

data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()

Complex database queries:

plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table

surveys # The plot_id column also features in the surveys table

# Join databases method 1
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()

NoSQL

nodbi: the NoSQL Database Connector

Github

R source

https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking 1000+ commits to look at daily changes.

If we are interested in a certain branch (say 3.2), look for R-3-2-branch.

R packages (only) source (metacran)

Bioconductor packages source

Announcement, https://github.com/Bioconductor-mirror

Send local repository to Github in R by using reports package

http://www.youtube.com/watch?v=WdOI_-aZV0Y

My collection

How to download

Clone ~ Download.

  • Command line
git clone https://gist.github.com/4484270.git

This will create a subdirectory called '4484270' with all cloned files there.

  • Within R
library(devtools)
source_gist("4484270")

or First download the json file from

https://api.github.com/users/MYUSERLOGIN/gists

and then

library(RJSONIO)
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})

Jekyll

An Easy Start with Jekyll, for R-Bloggers

Connect R with Arduino

Android App

Common plots tips

Create an empty plot

plot.new()

Overlay plots

How to Overlay Plots in R-Quick Guide with Example.

#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)

Save the par() and restore it

Example 1: Don't use old.par <- par() directly. no.readonly = FALSE by default. * The `no.readonly = TRUE` argument in the par() function in R is used to get the full list of graphical parameters that can be restored.

  • When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
  • If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)

Example 2: Use it inside a function with the on.exit(0 function.

ex <- function() {
   old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
   on.exit(par(old.par))
   ## ... do lots of par() settings and plots
   ## ...
   invisible() #-- now,  par(old.par)  will be executed
}

Example 3: It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.

ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar

Grouped boxplots

Weather Time Line

The plot looks similar to a boxplot though it is not. See a screenshot on Android by Sam Ruston.

Horizontal bar plot

library(ggplot2)
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA", 
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) + 
  coord_flip() + xlab("") + ylab("Fold Change")   

File:Ggplot2bar.svg

Include bar values in a barplot

Use text().

Or use geom_text() if we are using the ggplot2 package. See an example here or this.

For stacked barplot, see this post.

Grouped barplots

library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) + 
       geom_bar(position="dodge", stat="identity")
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042

Math expression

# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
     ylab = expression(phi^{rho + a}),
     main = "Pure Expressions")

# Superscript
plot(1:10, main = expression("My Title"^2)) 
# Subscript
plot(1:10, main = expression("My Title"[2]))  

# Expressions with Spacing
# '~' is to add space and '*' is to squish characters together
plot(1:10, xlab= expression(Delta * 'C'))
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
     ylab = expression(phi^{rho + a} * z * w),
     main = "Pure Expressions with Spacing")

# Expressions with Text
plot(x,y, 
     xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")), 
     ylab = expression(paste("Here is some text of ", phi^{rho})), 
     main = "Expressions with Text")

# Substituting Expressions
plot(x,y, 
     xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)), 
     ylab = substitute(paste("e is = ", e ), list(e = ee)), 
     main = "Substituted Expressions")

Impose a line to a scatter plot

  • abline + lsfit # least squares
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
  • abline + line # robust line fitting
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
  • lines
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')

How to actually make a quality scatterplot in R: axis(), mtext()

How to actually make a quality scatterplot in R

3D scatterplot

Rotating x axis labels for barplot

https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot

barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)

Set R plots x axis to show at y=0

https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0

plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")

Different colors of axis labels in barplot

See Vary colors of axis labels in R based on another variable

Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.

tN <- table(Ni <- stats::rpois(100, lambda = 5))
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")

Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.

barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)

Use text() to draw labels on X/Y-axis including rotation

par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
     adj = 1, labels = myData$names, xpd = TRUE)

Vertically stacked plots with the same x axis

https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in

Include labels on the top axis/margin: axis() and mtext()

plot(1:4, rnorm(4), axes = FALSE)
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3

See also 15_Questions_All_R_Users_Have_About_Plots

This can be used to annotate each plot with the script name, date, ...

mtext(text=paste("Prepared on", format(Sys.time(), "%d %B %Y at %H:%M")), 
      adj=.99,  # text align to right 
      cex=.75, side=3, las=1, line=2)

ggplot2 uses breaks instead of at parameter. See ggplot2 → Add axis on top or right hand side, ggplot2 → scale_x_continus(name, breaks, labels) and the scale_continuous documentation.

Legend tips

Add legend to a plot in R

Increase/decrease legend font size cex & ggplot2 package case.

plot(rnorm(100))
# op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)

legend inset. Default is 0. % (from 0 to 1) to draw the legend away from x and y axis. The inset argument with negative values moves the legend outside the plot.

legend("bottomright", inset=.05, )

legend without a box

legend(, bty = "n")

Add a legend title

legend(, title = "")

Add a common legend to multiple plots. Use the layout function.

Superimpose a density plot or any curves

Use lines().

Example 1

plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))

plot(density(x), col = "#6F69AC", lwd = 3)
lines(density(y), col = "#95DAC1", lwd = 3)
lines(density(z), col = "#FFEBA1", lwd = 3)

Example 2

require(survival)
n = 10000
beta1 = 2; beta2 = -1
lambdaT = 1 # baseline hazard
lambdaC = 2  # hazard of censoring
set.seed(1234)
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2)) 
C <- rweibull(n, shape=1, scale=lambdaC)   
time = pmin(T,C)  
status <- 1*(T <= C) 
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1), 
     ylab="Survival probability",
     main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull

Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".

log scale

If we set y-axis to use log-scale, then what we display is the value log(Y) or log10(Y) though we still label the values using the input. For example, when we plot c(1, 10, 100) using the log scale, it is like we draw log10(c(1, 10, 100)) = c(0,1,2) on the plot but label the axis using the true values c(1, 10, 100).

File:Logscale.png

Custom scales

Using custom scales with the 'scales' package

Time series

Time series stock price plot

library(quantmod)
getSymbols("AAPL")
getSymbols("IBM") # similar to AAPL
getSymbols("CSCO") # much smaller than AAPL, IBM
getSymbols("DJI") # Dow Jones, huge 
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)", 
    col='orange', subset = '2017::2017-08')

tail(Cl(DJI))

tidyquant: Getting stock data

The 'largest stock profit or loss' puzzle: efficient computation in R

Timeline plot

Clockify

Clockify

Circular plot

Word cloud

Text mining

World map

Visualising SSH attacks with R (rworldmap and rgeolocate packages)

Diagram/flowchart/Directed acyclic diagrams (DAGs)

DiagrammeR

diagram

Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams

DAGitty (browser-based and R package)

dagR

Gmisc

Easiest flowcharts eveR?

Concept Maps

concept-maps where the diagrams are generated from https://app.diagrams.net/.

flow

flow, How To Draw Flow Diagrams In R

Venn Diagram

Venn diagram

hexbin plot

Bump chart/Metro map

https://dominikkoch.github.io/Bump-Chart/

Amazing/special plots

See Amazing plot.

Google Analytics

GAR package

http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html

Linear Programming

http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/

Linear Algebra

Amazon Alexa

R and Singularity

https://rviews.rstudio.com/2017/03/29/r-and-singularity/

Teach kids about R with Minecraft

http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html

Secure API keys

Securely store API keys in R scripts with the "secret" package

Credentials and secrets

How to manage credentials and secrets safely in R

Hide a password

keyring package

getPass

getPass

Vision and image recognition

Creating a Dataset from an Image

Creating a Dataset from an Image in R Markdown using reticulate

Turn pictures into coloring pages

https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0

Numerical optimization

CRAN Task View: Numerical Mathematics, CRAN Task View: Optimization and Mathematical Programming

Ryacas: R Interface to the 'Yacas' Computer Algebra System

Doing Maths Symbolically: R as a Computer Algebra System (CAS)

Game

Music

  • gm. Require to install MuseScore, an open source and free notation software.

SAS

sasMap Static code analysis for SAS scripts

R packages

R packages

Tricks

Getting help

Better Coder/coding, best practices

E-notation

6.022E23 (or 6.022e23) is equivalent to 6.022×10^23

Getting user's home directory

See What are HOME and working directories?

# Windows
normalizePath("~")   # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")   # "C:/Users/brb/Documents"

# Mac
normalizePath("~")   # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # "/Users/brb"

# Linux
normalizePath("~")   # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # [1] "/home/brb"

tempdir()

The path is a per-session temporary directory. On parallel use, R processes forked by functions such as mclapply and makeForkCluster in package parallel share a per-session temporary directory.

Distinguish Windows and Linux/Mac, R.Version()

identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.

get_os <- function(){
  sysinf <- Sys.info()
  if (!is.null(sysinf)){
    os <- sysinf['sysname']
    if (os == 'Darwin')
      os <- "osx"
  } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
names(R.Version())
#  [1] "platform"       "arch"           "os"             "system"        
#  [5] "status"         "major"          "minor"          "year"          
#  [9] "month"          "day"            "svn rev"        "language"      
# [13] "version.string" "nickname" 
getRversion()
# [1] ‘4.3.0’

Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only)

If we like to install R packages to a personal directory, follow this. Just add the line

R_LIBS_SITE=F:/R/library

to the file R_HOME/etc/x64/Renviron.site. In R, run Sys.getenv("R_LIBS_SITE") or Sys.getenv("R_LIBS_USER") to query the environment variable. See Environment Variables.

What is the best place to save Rconsole on Windows platform

Put/create the file <Rconsole> under C:/Users/USERNAME/Documents folder so no matter how R was upgraded/downgraded, it always find my preference.

My preferred settings:

  • Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
  • Size: 12
  • background: black
  • normaltext: white
  • usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow

and others (default options)

  • pagebg: white
  • pagetext: navy
  • highlight: DarkRed
  • dataeditbg: white
  • dataedittext: navy (View() function)
  • dataedituser: red
  • editorbg: white (edit() function)
  • editortext: black

A copy of the Rconsole is saved in github.

How R starts up

https://rstats.wtf/r-startup.html

startup - Friendly R Startup Configuration

https://github.com/henrikbengtsson/startup

Saving and loading history automatically: .Rprofile & local()

  • savehistory("filename"). It will save everything from the beginning to the command savehistory() to a text file.
  • .Rprofile will automatically be loaded when R has started from that directory
  • Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See Project-oriented workflow.
  • .Rprofile has been created/used by the packrat package to restore a packrat environment. See the packrat/init.R file and R packages → packrat.
  • Customizing Startup from R in Action, Fun with .Rprofile and customizing R startup
    • You can also place a .Rprofile file in any directory that you are going to run R from or in the user home directory.
    • At startup, R will source the Rprofile.site file. It will then look for a .Rprofile file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
    options(continue="  ") # default is "+ "
    options(prompt="R> ", continue=" ")
    options(editor="nano") # default is "vi" on Linux
    # options(htmlhelp=TRUE) 
    
    local({r <- getOption("repos")
          r["CRAN"] <- "https://cran.rstudio.com"
          options(repos=r)})
    
    .First <- function(){
     # library(tidyverse)
     cat("\nWelcome at", date(), "\n")
    }
    
    .Last <- function(){
     cat("\nGoodbye at ", date(), "\n")
    }  
    
  • https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
  • The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define R_HISTFILE system variable.
  • local() function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)

Linux or Mac

In ~/.profile or ~/.bashrc I put:

export R_HISTFILE=~/.Rhistory

In ~/.Rprofile I put:

if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory")) 
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Windows

If you launch R by clicking its icon from Windows Desktop, the R starts in C:\User\$USER\Documents directory. So we can create a new file .Rprofile in this directory.

if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Disable "Save workspace image?" prompt when exit R?

How to disable "Save workspace image?" prompt in R?

R release versions

rversions: Query the main 'R' 'SVN' repository to find the released versions & dates.

getRversion()

getRversion()
[1] ‘4.3.0’

Detect number of running R instances in Windows

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
INFO: No tasks are running which match the specified criteria.

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rgui.exe                      1096 Console                    1     44,712 K

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rserve.exe                    6108 Console                    1    381,796 K

In R, we can use

> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
[1] ""                                                                            
[2] "Image Name                     PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1     44,804 K"

> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3

Editor

http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs

  • Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
    • Edit the file C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el with something like
    (setq-default inferior-R-program-name
                  "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
    

GUI for Data Analysis

Update to Data Science Software Popularity 6/7/2023

BlueSky Statistics

Rcmdr

http://cran.r-project.org/web/packages/Rcmdr/index.html. After loading a dataset, click Statistics -> Fit models. Then select Linear regression, Linear model, GLM, Multinomial logit model, Ordinal regression model, Linear mixed model, and Generalized linear mixed model. However, Rcmdr does not include, e.g. random forest, SVM, glmnet, et al.

Deducer

http://cran.r-project.org/web/packages/Deducer/index.html

jamovi

Scope

See

source()

## foo.R ##
cat(ArrayTools, "\n")
## End of foo.R

# 1. Error
predict <- function() {
  ArrayTools <- "C:/Program Files" # or through load() function 
  source("foo.R")                  # or through a function call; foo()
}
predict()   # Object ArrayTools not found

# 2. OK. Make the variable global
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict()  
ArrayTools

# 3. OK. Create a global variable
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()

Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.

Example 1.

> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a      N5
2    b     999
3    c     999
4    d     999
5    e     999
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999

Example 2. How can we set global variables inside a function? The answer is to use the "<<-" operator or assign(, , envir = .GlobalEnv) function.

Other resource: Advanced R by Hadley Wickham.

Example 3. Writing functions in R, keeping scoping in mind

New environment

Run the same function on a bunch of R objects

mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(mye[[n]])

Just look at the contents of rda file without saving to anywhere (?load)

local({
   load("myfile.rda")
   ls()
})

Or use attach() which is a wrapper of load(). It creates an environment and slots it into the list right after the global environment, then populates it with the objects we're attaching.

attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
ls(pos = 2)
##  also typically need to cleanup the search path:
detach("file:all.rda")

If we want to read data from internet, load() works but not attach().

con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840

source() case.

myEnv <- new.env()    
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))

str( , max) function

Use max.level parameter to avoid a long display of the structure of a complex R object. Use give.head = FALSE to hide the attributes. See ?str

If we use str() on a function like str(lm), it is equivalent to args(lm)

For a complicated list object, it is useful to use the max.level argument; e.g. str(, max.level = 1)

For a large data frame, we can use the tibble() function; e.g. mydf %>% tibble()

tidy() function

broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See here.

View all objects present in a package, ls()

https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get

> devtools::load_all("mypackage")
> search()
 [1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
 [4] "package:stats"     "package:graphics"  "package:grDevices"
 [7] "package:utils"     "package:datasets"  "package:methods"
[10] "Autoloads"         "package:base"

> ls("package:mypackage")
[1] "_mypackage_rcpp_hello_world" "evalCpp"                     "library.dynam.unload"       
[4] "rcpp_hello_world"            "system.file"

Note that the first argument of ls() (or detach()) is used to specify the environment. It can be

  • an integer (the position in the ‘search’ list);
  • the character string name of an element in the search list;
  • an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).

Speedup R code

Profiler

&& vs &

See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.

  • The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
  • The longer form evaluates left to right examining only the first element of each vector. The return is one value.
  • The longer form evaluates left to right examining only the first element of each vector. Evaluation proceeds only until the result is determined.
  • The idea of the longer form && in R seems to be the same as the && operator in linux shell; see here.
  • Single or double?: AND operator and OR operator in R. The confusion might come from the inconsistency when choosing these operators in different languages. For example, in C, & performs bitwise AND, while && does Boolean logical AND.
  • Think of && as a stricter &
c(T,F,T) & c(T,T,T)
# [1]  TRUE FALSE  TRUE
c(T,F,T) && c(T,T,T)
# [1] TRUE
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated

It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.

> foo <- function(arg, L) {
   # Suppose 'L' is meaningful only if 'arg' is provided
   # 
   # Evaluate 'L' only if 'arg' is provided
   #
   if (!missing(arg) && L) {
     print("L is true")
   } else {
     print("Either arg is missing or L is FALSE")
   }
 }
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"

Other examples: && is more flexible than &.

nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))

if (!is.null(exprTest) && any(is.na(exprTest))) { ... }

for-loop, control flow

Vectorization

sapply vs vectorization

Speed test: sapply vs vectorization

lapply vs for loop

split() and sapply()

split() can be used to split a vector, columns or rows. See How to split a data frame?

  • Split divides the data in the vector or data frame x into the groups defined by f. The syntax is
split(x, f, drop = FALSE, …)
  • Split a vector into chunks. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the split() + lapply() + do.call() or split() + sapply() operations.
d <- 1:10
chunksize <- 4
ceiling(1:10/4)
# [1] 1 1 1 1 2 2 2 2 3 3
split(d, ceiling(seq_along(d)/chunksize))
# $`1`
# [1] 1 2 3 4
#
# $`2`
# [1] 5 6 7 8
#
# $`3`
# [1]  9 10
do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) ) 
#  1  2  3 
# 10 26 19

# bigmemory vignette
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                     function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
  • Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
split(mtcars,mtcars$cyl)

split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))   # split.data.frame() works for matrices
  • Split columns of a data frame/matrix.
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
  • split() + sapply() to merge columns. See below Mean of duplicated columns for more detail.
  • split() + sapply() to split a vector. See nsFilter() function which can remove duplicated probesets/rows using unique Entrez Gene IDs (genefilter package). The source code of nsFilter() and findLargest().
tSsp = split.default(testStat, lls) 
# testStat is a vector of numerics including probeset IDs as names
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.

sapply(tSsp, function(x) names(which.max(x))) 
# return a vector of probset IDs of length of unique entrez IDs

strsplit and sapply

> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
> strsplit(namedf, " ")
1
[1] "John" "ABC" 

2
[1] "Mary" "CDE" 

3
[1] "Kat" "FGH"

> sapply(strsplit(namedf, " "), "[", 1)
[1] "John" "Mary" "Kat" 
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"

Mean of duplicated columns: rowMeans; compute Means by each row

  • Reduce columns of a matrix by a function in R. To use rowMedians() instead of rowMeans(), we need to install matrixStats from CRAN.
    set.seed(1)
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
    colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
    res <- sapply(split(1:ncol(x), colnames(x)), 
                  function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
    res  # notice the sorting of columns
           a  b  c
     [1,] 46  1 31
     [2,] 47 12 32
     [3,] 48 13 33
     [4,] 49 14 34
     [5,] 50 15 35
     [6,] 51 16 36
     [7,] 52 17 37
     [8,] 53 18 38
     [9,] 54 19 39
    [10,] 55 20 40
    
    # vapply() is safter than sapply(). 
    # The 3rd arg in vapply() is a template of the return value.
    res2 <- vapply(split(1:ncol(x), colnames(x)), 
                   function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
                   rep(0, nrow(x)))
  • colSums, rowSums, colMeans, rowMeans (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster.
    rowMeans(x, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
    apply(x, 1, mean, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
  • matrixStats: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
  • From for() loops to the split-apply-combine paradigm for column-wise tasks: the transition for a dinosaur

Mean of duplicated rows: colMeans and rowsum

  • colMeans(x, na.rm = FALSE, dims = 1), take mean per columns & sum over rows. It returns a vector. Other similar idea functions include colSums, rowSums, rowMeans.
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
    rownames(x) <- c(rep("b", 2), rep("c", 3), rep("d", 4), "a") # move 'a' to the last
    res <- sapply(split(1:nrow(x), rownames(x)), 
                  function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
    res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
    res  # still a matrix, rows are ordered
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5 12.0 22.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    table(rownames(x))
    # a b c d
    # 1 2 3 4
    
    aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
    #   Group.1   V1   V2   V3   V4   V5   V6
    # 1       a 10.0 20.0 30.0 40.0 50.0 60.0
    # 2       b  1.5 12.0 22.0 31.5 41.5 51.5
    # 3       c  4.0 14.0 24.0 34.0 44.0 54.0
    # 4       d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays
  • rowsum(x, group, reorder = TRUE, …). Sum over rows. It returns a matrix. This is very special. It's not the same as rowSums. There is no "colsum" function. It has the speed advantage over sapply+colSums OR aggregate.
    group <- rownames(x)
    rowsum(x, group, na.rm=T)/as.vector(table(group))
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5  6.0 11.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • by() function. Calculating change from baseline in R
  • See aggregate Function in R- A powerful tool for data frames & summarize in r, Data Summarization In R
  • aggregate() function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. Don't use aggregate post.
    > attach(mtcars)
    dim(mtcars)
    [1] 32 11
    > head(mtcars)
                       mpg cyl disp  hp drat    wt  qsec vs am gear carb
    Mazda RX4         21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
    Mazda RX4 Wag     21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
    Datsun 710        22.8   4  108  93 3.85 2.320 18.61  1  1    4    1
    Hornet 4 Drive    21.4   6  258 110 3.08 3.215 19.44  1  0    3    1
    Hornet Sportabout 18.7   8  360 175 3.15 3.440 17.02  0  0    3    2
    Valiant           18.1   6  225 105 2.76 3.460 20.22  1  0    3    1
    > with(mtcars, table(cyl, vs))
       vs
    cyl  0  1
      4  1 10
      6  3  4
      8 14  0
    > aggdata <-aggregate(mtcars, by=list(cyl,vs),  FUN=mean, na.rm=TRUE)
    > print(aggdata)
      Group.1 Group.2      mpg cyl   disp       hp     drat       wt     qsec vs
    1       4       0 26.00000   4 120.30  91.0000 4.430000 2.140000 16.70000  0
    2       6       0 20.56667   6 155.00 131.6667 3.806667 2.755000 16.32667  0
    3       8       0 15.10000   8 353.10 209.2143 3.229286 3.999214 16.77214  0
    4       4       1 26.73000   4 103.62  81.8000 4.035000 2.300300 19.38100  1
    5       6       1 19.12500   6 204.55 115.2500 3.420000 3.388750 19.21500  1
             am     gear     carb
    1 1.0000000 5.000000 2.000000
    2 1.0000000 4.333333 4.666667
    3 0.1428571 3.285714 3.500000
    4 0.7000000 4.000000 1.500000
    5 0.0000000 3.500000 2.500000
    > detach(mtcars)
    
    # Another example: select rows with a minimum value from a certain column (yval in this case)
    > mydf <- read.table(header=T, text='
     id xval yval
     A 1  1
     A -2  2
     B 3  3
     B 4  4
     C 5  5
     ')
    > x = mydf$xval
    > y = mydf$yval
    > aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
      id xval yval
    1  A    1    1
    2  B    3    3
    3  C    5    5
    

Mean by Group

Mean by Group in R (2 Examples) | dplyr Package vs. Base R

aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                           # Specify function (i.e. mean)
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                         # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
               list(name = mean))               # Specify function
  • ave(x, ..., FUN),
  • aggregate(x, by, FUN),
  • by(x, INDICES, FUN): return is a list
  • tapply(): return results as a matrix or array. Useful for ragged array.

Apply family

Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.

The following list gives a hierarchical relationship among these functions.

  • apply(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
  • lapply(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
    • sapply(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
      • replicate(n, expr, simplify = "array")
    • mapply(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
      • Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
      • Map(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
    • vapply(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
    • rapply(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
  • tapply(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a "Ragged" Array. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
    • aggregate(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each columns of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame. This is used to summarize data.
    • by(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each subset data frame split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
  • eapply(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment

Difference between apply vs sapply vs lapply vs tapply?

  • apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
  • lapply - When you want to apply a function to each element of a list in turn and get a list back.
  • sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
  • tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.

Some short examples:

Apply vs for loop

Note that, apply's performance is not always better than a for loop. See

Progress bar

What is the cost of a progress bar in R?

The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out this post. It uses winProgressBar() and setWinProgressBar() functions.

e-Rum 2020 Slides on Progressr by Henrik Bengtsson. progressr 0.8.0: RStudio's progress bar, Shiny progress updates, and absolute progress, progressr 0.10.1: Plyr Now Supports Progress Updates also in Parallel

simplify option in sapply()

library(KEGGREST)

names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2

names3 <- keggGet(c("hsa05340"))
names4 <- sapply(names3, function(x) x$GENE)
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify 
# [1] 1

lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists

  • mapply() documentation. Use mapply() to merge lists.
    mapply(rep, 1:4, 4:1)
    mapply(rep, times = 1:4, x = 4:1)
    mapply(function(x, y) seq_len(x) + y,
           c(a =  1, b = 2, c = 3),  # names from first
           c(A = 10, B = 0, C = -10))
    mapply(c, firstList, secondList, SIMPLIFY=FALSE)
    
  • Finding the Expected value of the maximum of two Bivariate Normal variables with simulation sapply + mapply.
    z <- mapply(function(u, v) { max(u, v) }, 
                u = x[, 1], v = x[, 2])
    
  • Map() and Reduce() in functional programming
  • Map(), Reduce(), and Filter() from Advanced R by Hadley
    • If you have two or more lists (or data frames) that you need to process in parallel, use Map(). One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to mapply() function and is more concise than lapply(). Advanced R has a comment that Map() is better than mapply().
      # Syntax: Map(f, ...)
      
      xs <- replicate(5, runif(10), simplify = FALSE)
      ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
      Map(weighted.mean, xs, ws)
      
      # instead of a more clumsy way
      lapply(seq_along(xs), function(i) {
        weighted.mean(xsi, wsi)
      })
      
    • Reduce() reduces a vector, x, to a single value by recursively calling a function, f, two arguments at a time. A good example of using Reduce() function is to read a list of matrix files and merge them. See How to combine multiple matrix frames into one using R?
      # Syntax: Reduce(f, x, ...)
      
      > m1 <- data.frame(id=letters[1:4], val=1:4)
      > m2 <- data.frame(id=letters[2:6], val=2:6)
      > merge(m1, m2, "id", all = T)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      > m <- list(m1, m2)
      > Reduce(function(x,y) merge(x,y, "id",all=T), m)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      

sapply & vapply

See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up this example.

rapply - recursive version of lapply

replicate

https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r

> replicate(5, rnorm(3))
           [,1]       [,2]       [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477

See parSapply() for a parallel version of replicate().

Vectorize

> rep(1:4, 4:1)
 [1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
1
[1] 1 1 1 1

2
[1] 2 2 2

3
[1] 3 3

4
[1] 4
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2)) 
[1] 1.6491761 0.9610109
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15

myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
myfunc2(1, 2) # 2 
myfunc2(3, 5) # 15
myfunc2(c(1,3), c(2,5)) # NA
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used

plyr and dplyr packages

Practical Data Science for Stats - a PeerJ Collection

The Split-Apply-Combine Strategy for Data Analysis (plyr package) in J. Stat Software.

A quick introduction to plyr with a summary of apply functions in R and compare them with functions in plyr package.

  1. plyr has a common syntax -- easier to remember
  2. plyr requires less code since it takes care of the input and output format
  3. plyr can easily be run in parallel -- faster

Tutorials

Examples of using dplyr:

tibble

Tibbles are data frames, but slightly tweaked to work better in the tidyverse.

Tibble objects

  • it does not have row names (cf data frame),
  • it never changes the type of the inputs (e.g. it never converts strings to factors!),
  • it never changes the names of variables

Tibbles Vignette

> data(pew, package = "efficient")
> dim(pew) 
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"     "tbl"        "data.frame"

> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
# A tibble: 162 x 3
                                                       religion Income Count
                                                          <chr>  <chr> <int>
 1                                                     Agnostic  <$10k    27
 2                                                      Atheist  <$10k    12
 ...
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
[1] NA
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count,  :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)3)
[1] 181.6975

To show all rows of a tibble object, use the print() method.

print(tbObj, n= Inf)

tbObj %>% print(n= nrow(.))

If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.

Subsetting: to extract a column from a tibble object, use [[ or $ or dplyr::pull(). Select Data Frame Columns in R.

TibbleObject$VarName
# OR
TibbleObject"VarName"
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore

dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# OR
dplyr::select(TibbleObject, 2:5) # 

Convert a data frame to a tibble See Tibble Data Format in R: Best and Modern Way to Work with Your Data

my_data <- as_tibble(iris)
class(my_data)

To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)

llply()

llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.

LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])

where rLLID is a list of entrez ID. For example,

get("org.Hs.egGO")[["6772"]]

returns a list of 49 GOs.

ddply()

http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html

ldply()

An R Script to Automatically download PubMed Citation Counts By Year of Publication

Performance/speed comparison

Performance comparison of converting list to data.frame with R language

Using R's set.seed() to set seeds for use in C/C++ (including Rcpp)

http://rorynolan.rbind.io/2018/09/30/rcsetseed/

get_seed()

See the same blog

get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}

Note: .Machine$integer.max = 2147483647 = 2^31 - 1.

Random seeds

By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See ?Random.

set.seed(as.numeric(Sys.time()))

set.seed(as.numeric(Sys.Date()))  # same seed for each day

.Machine and the largest integer, double

See ?.Machine.

                          Linux/Mac  32-bit Windows 64-bit Windows
double.eps              2.220446e-16   2.220446e-16   2.220446e-16
double.neg.eps          1.110223e-16   1.110223e-16   1.110223e-16
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
double.base             2.000000e+00   2.000000e+00   2.000000e+00
double.digits           5.300000e+01   5.300000e+01   5.300000e+01
double.rounding         5.000000e+00   5.000000e+00   5.000000e+00
double.guard            0.000000e+00   0.000000e+00   0.000000e+00
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
double.exponent         1.100000e+01   1.100000e+01   1.100000e+01
double.min.exp         -1.022000e+03  -1.022000e+03  -1.022000e+03
double.max.exp          1.024000e+03   1.024000e+03   1.024000e+03
integer.max             2.147484e+09   2.147484e+09   2.147484e+09
sizeof.long             8.000000e+00   4.000000e+00   4.000000e+00
sizeof.longlong         8.000000e+00   8.000000e+00   8.000000e+00
sizeof.longdouble       1.600000e+01   1.200000e+01   1.600000e+01
sizeof.pointer          8.000000e+00   4.000000e+00   8.000000e+00

NA when overflow

tmp <- 156287L
tmp*tmp
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647

How to select a seed for simulation or randomization

set.seed() allow alphanumeric seeds

https://stackoverflow.com/a/10913336

set.seed(), for loop and saving random seeds

  • Detect When the Random Number Generator Was Used
    if (interactive()) {
      invisible(addTaskCallback(local({
        last <- .GlobalEnv$.Random.seed
        
        function(...) {
          curr <- .GlobalEnv$.Random.seed
          if (!identical(curr, last)) {
            msg <- "NOTE: .Random.seed changed"
            if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
            message(msg)
            last <<- curr
          }
          TRUE
        }
      }), name = "RNG tracker"))
    }
    
  • http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
    set.seed(1001) 
    data <- vector("list", 30) 
    seeds <- vector("list", 30) 
    for(i in 1:30) { 
      seeds[[i]] <- .Random.seed 
      data[[i]] <- runif(5) 
    } 
     
    # If we save and load .Random.seed from a file using scan(), make
    # sure to convert its type from doubles to integers.
    # Otherwise, .Random.seed will complain!
    
    .Random.seed <- seeds[[23]]  # restore 
    data.23 <- runif(5) 
    data.23 
    data[[23]] 
    
  • impute.knn
  • Duncan Murdoch: This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed. See ?.Random.seed for details.
  • Uwe Ligges's comment: set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.
  • Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
  • Local randomness in R.

sample()

sample() inaccurate on very large populations, fixed in R 3.6.0

# R 3.5.3
set.seed(123)
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1 
# 400070 599930 
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

# R 3.6.0
# docker run --net=host -it --rm r-base:3.6.0
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

Getting different results with set.seed() in RStudio

Getting different results with set.seed(). It's possible that you're loading an R package that is changing the requested random number generator; RNGkind().

dplyr::sample_n()

The function has a parameter weight. For example if we have some download statistics for each day and we want to do sampling based on their download numbers, we can use this function.

Regular Expression

See here.

Read rrd file

on.exit()

Examples of using on.exit(). In all these examples, add = TRUE is used in the on.exit() call to ensure that each exit action is added to the list of actions to be performed when the function exits, rather than replacing the previous actions.

  • Database connections
    library(RSQLite)
    sqlite_get_query <- function(db, sql) {
      conn <- dbConnect(RSQLite::SQLite(), db)
      on.exit(dbDisconnect(conn), add = TRUE)
      dbGetQuery(conn, sql)
    }
    
  • File connections
    read_chars <- function(file_name) {
      conn <- file(file_name, "r")
      on.exit(close(conn), add = TRUE)
      readChar(conn, file.info(file_name)$size)
    }
    
  • Temporary files
    history_lines <- function() {
      f <- tempfile()
      on.exit(unlink(f), add = TRUE)
      savehistory(f)
      readLines(f, encoding = "UTF-8")
    }
    
  • Printing messages
    myfun = function(x) {
      on.exit(print("first"))
      on.exit(print("second"), add = TRUE)
      return(x)
    }
    

file, connection

  • cat() and scan() (read data into a vector or list from the console or file)
  • read() and write()
  • read.table() and write.table()
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)

foo <- function() {
  con <- file()
  ...
  on.exit(close(con))
  ...
}

Error in close.connection(f) : invalid connection. If we want to use close(con), we have to specify how to open the connection; such as

con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)

withr package

https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by languageserver.

Clipboard (?connections), textConnection(), pipe()

  • On Windows, we can use readClipboard() and writeClipboard().
    source("clipboard")
    read.table("clipboard")
    
  • Clipboard -> R. Reading/writing clipboard on macOS. Use textConnection() function:
    x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
    # Or on Mac
    x <- read.delim(pipe("pbpaste"))
    # safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
    

    An example is to copy data from this post. In this case we need to use read.table() instead of read.delim().

  • R -> clipboard on Mac. Note: pbcopy and pbpaste are macOS terminal commands. See pbcopy & pbpaste: Manipulating the Clipboard from the Command Line.
    • pbcopy: takes standard input and places it in the clipboard buffer
    • pbpaste: takes data from the clipboard buffer and writes it to the standard output
    clip <- pipe("pbcopy", "w")
    write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
    # write.table(data.frame(Var1, Var2), file = clip, row.names=F, quote=F, sep="\t")
    close(clip)
    
  • Clipboard -> Excel.
    • Method 1: Paste icon -> Text import wizard -> Delimit (Tab, uncheck Space) or Fixed width depending on the situation -> Finish.
    • Method 2: Ctrl+v first. Then choose Data -> Text to Columns. Fixed width -> Next -> Next -> Finish.
  • On Linux, we need to install "xclip". See R Copy from Clipboard in Ubuntu Linux. It seems to work.
    # sudo apt-get install xclip
    read.table(pipe("xclip -selection clipboard -o",open="r"))
    

clipr

clipr: Read and Write from the System Clipboard

read/manipulate binary data

  • x <- readBin(fn, raw(), file.info(fn)$size)
  • rawToChar(x[1:16])
  • See Biostrings C API

String Manipulation

format(): padding with zero

ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"

noquote()

noqute Print character strings without quotes.

stringr package

glue package

  • glue. Useful in a loop and some function like ggtitle() or ggsave().
    library(glue)
    name <- "Fred"
    glue('My name is {name}.')  # My name is Fred.
    
  • String interpolation

Raw data type

Fun with strings, Cyrillic alphabets

a1 <- "А"
a2 <- "A"
a1 == a2
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41

number of characters limit

It's a limit on a (single) input line in the REPL

HTTPs connection

HTTPS connection becomes default in R 3.2.2. See

R 3.3.2 patched The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)

setInternet2

There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.

Read the discussion reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in this post. The following demonstrated the original problem.

url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)

It seems the bug was fixed in R 3.2-branch. See 8/16/2015 patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to InternetOpenUrl() function of wininet library. This article and this post explain differences of active and passive FTP.

The following R command will show the exact svn revision for the R you are currently using.

R.Version()$"svn rev"

If setInternet2(T), then https protocol is supported in download.file().

When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.

The setInternet2() function is defined in R> src> library> utils > R > windows > sysutils.R.

R up to 3.2.2

setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))

See also

  • <src/include/Internal.h> (declare do_setInternet2()),
  • <src/main/names.c> (show do_setInternet2() in C)
  • <src/main/internet.c> (define do_setInternet2() in C).

Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file. If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2, it is also possible to download from https by setIneternet2(T).

R 3.3.0

setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}

Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.

Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan()

In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, Inf and NaN as input or output.

See ?is.finite.

How to replace Inf with NA in All or Specific Columns of the Data Frame

replace() function

File/path operations

  • list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
  • file.info()
  • dir.create()
  • file.create()
  • file.copy()
  • file.exists()
  • basename() - remove the parent path, dirname() - returns the part of the path up to but excluding the last path separator
    > file.path("~", "Downloads")
    [1] "~/Downloads"
    > dirname(file.path("~", "Downloads"))
    [1] "/home/brb"
    > basename(file.path("~", "Downloads"))
    [1] "Downloads"
    
  • path.expand("~/.Renviron") # "/home/brb/.Renviron"
  • normalizePath() # Express File Paths in Canonical Form
    > cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
    /usr/lib/R
    /tmp/RtmpzvDhAe
    
  • system.file() - Finds the full file names of files in packages etc
    > system.file("extdata", "ex1.bam", package="Rsamtools")
    [1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
    

read/download/source a file from internet

Simple text file http

retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)

Zip, RData, gz file and url() function

x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)

Here url() function is like file(), gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.

Another example is Read gzipped csv directly from a url in R

con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
                       "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))

Another example of using url() is

load(url("http:/www.example.com/example.RData"))

This does not work with load(), dget(), read.table() for files on OneDrive. In fact, I cannot use wget with shared files from OneDrive. The following trick works: How to configure a OneDrive file for use with wget.

Dropbox is easy and works for load(), wget, ...

R download .RData or Directly loading .RData from github from Github.

zip function

This will include 'hallmarkFiles' root folder in the files inside zip.

zip(zipfile = 'myFile.zip', 
    files = dir('hallmarkFiles', full.names = TRUE))

# Verify/view the files. 'list = TRUE' won't extract 
unzip('testZip.zip', list = TRUE) 

downloader package

This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.

Google drive file based on https using RCurl package

require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))

Google sheet file using googlesheets package

Reading data from google sheets into R

Github files https using RCurl package

x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt", 
            ssl.verifypeer = FALSE)
read.table(text=x)

data summary table

summarytools: create summary tables for vectors and data frames

https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.

skimr: A frictionless, pipeable approach to dealing with summary statistics

skimr for useful and tidy summary statistics

modelsummary

modelsummary: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready

broom

Tidyverse->broom

Create publication tables using tables package

See p13 for example at here

R's tables packages is the best solution. For example,

> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                  
                Sepal.Length      Sepal.Width     
 Species    n   mean         sd   mean        sd  
 setosa      50 5.01         0.35 3.43        0.38
 versicolor  50 5.94         0.52 2.77        0.31
 virginica   50 6.59         0.64 2.97        0.32
 All        150 5.84         0.83 3.06        0.44
> str(iris)
'data.frame':   150 obs. of  5 variables:
 $ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
 $ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
 $ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
 $ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
 $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...

and

# This example shows some of the less common options         
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
> z <- rnorm(100)+5
> fmt <- function(x) {
  s <- format(x, digits=2)
  even <- ((1:length(s)) %% 2) == 0
  s[even] <- sprintf("(%s)", s[even])
  s
}
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
                  Status              
 Sex    Statistic high   low    medium
 Female mean       4.88   4.96   5.17 
        sd        (1.20) (0.82) (1.35)
 Male   mean       4.45   4.31   5.05 
        sd        (1.01) (0.93) (0.75)

fgsea example

vignette & source code

(archived) ClinReport: Statistical Reporting in Clinical Trials

https://cran.r-project.org/web/packages/ClinReport/index.html

Append figures to PDF files

How to append a plot to an existing pdf file. Hint: use the recordPlot() function.

Save base graphics as pseudo-objects

Save base graphics as pseudo-objects in R. Note there are some cons with this approach.

pdf(NULL)
dev.control(displaylist="enable")
plot(df$x, df$y)
text(40, 0, "Random")
text(60, 2, "Text")
lines(stats::lowess(df$x, df$y))
p1.base <- recordPlot()
invisible(dev.off())

# Display the saved plot
grid::grid.newpage()
p1.base

Extracting tables from PDFs

Print tables

addmargins()

tableone

Some examples

Cox models

finalfit package

summary_factorlist() from the finalfit package.

table1

gtsummary

gt

Introduction to Clinical Tables with the {gt} Package

dplyr

https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.

tables::tabular()

gmodels::CrossTable()

https://www.statmethods.net/stats/frequencies.html

base::prop.table(x, margin)

New function ‘proportions()’ and ‘marginSums()’. These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’. for R 4.0.0.

R> m <- matrix(1:4, 2)
R> prop.table(m, 1) # row percentage
          [,1]      [,2]
[1,] 0.2500000 0.7500000
[2,] 0.3333333 0.6666667
R> prop.table(m, 2) # column percentage
          [,1]      [,2]
[1,] 0.3333333 0.4285714
[2,] 0.6666667 0.5714286

stats::xtabs()

stats::ftable()

> ftable(Titanic, row.vars = 1:3)
                   Survived  No Yes
Class Sex    Age                   
1st   Male   Child            0   5
             Adult          118  57
      Female Child            0   1
             Adult            4 140
2nd   Male   Child            0  11
             Adult          154  14
      Female Child            0  13
             Adult           13  80
3rd   Male   Child           35  13
             Adult          387  75
      Female Child           17  14
             Adult           89  76
Crew  Male   Child            0   0
             Adult          670 192
      Female Child            0   0
             Adult            3  20
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
             Survived  No Yes
Class Sex                    
1st   Male            118  62
      Female            4 141
2nd   Male            154  25
      Female           13  93
3rd   Male            422  88
      Female          106  90
Crew  Male            670 192
      Female            3  20
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
             Survived  No Yes
Sex    Class                 
Male   1st            118  62
       2nd            154  25
       3rd            422  88
       Crew           670 192
Female 1st              4 141
       2nd             13  93
       3rd            106  90
       Crew             3  20
> str(Titanic)
 table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
 - attr(*, "dimnames")=List of 4
  ..$ Class   : chr [1:4] "1st" "2nd" "3rd" "Crew"
  ..$ Sex     : chr [1:2] "Male" "Female"
  ..$ Age     : chr [1:2] "Child" "Adult"
  ..$ Survived: chr [1:2] "No" "Yes"
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
> x
          gear  3  4  5
cyl vs am              
4   0  0        0  0  0
       1        0  0  1
    1  0        1  2  0
       1        0  6  1
6   0  0        0  0  0
       1        0  2  1
    1  0        2  2  0
       1        0  0  0
8   0  0       12  0  0
       1        0  0  2
    1  0        0  0  0
       1        0  0  0
> ftable(x, row.vars = c(2, 4))
        cyl  4     6     8   
        am   0  1  0  1  0  1
vs gear                      
0  3         0  0  0  0 12  0
   4         0  0  0  2  0  0
   5         0  1  0  1  0  2
1  3         1  0  2  0  0  0
   4         2  6  2  0  0  0
   5         0  1  0  0  0  0
> 
> ## Start with expressions, use table()'s "dnn" to change labels
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
         dnn = c("Cylinders", "V/S", "Transmission", "Gears"))

          Cylinders     4     6     8   
          Transmission  0  1  0  1  0  1
V/S Gears                               
0   3                   0  0  0  0 12  0
    4                   0  0  0  2  0  0
    5                   0  1  0  1  0  2
1   3                   1  0  2  0  0  0
    4                   2  6  2  0  0  0
    5                   0  1  0  0  0  0

tracemem, data type, copy

How to avoid copying a long vector

Tell if the current R is running in 32-bit or 64-bit mode

8 * .Machine$sizeof.pointer

where sizeof.pointer returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.

32- and 64-bit

See R-admin.html.

  • For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
  • Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
  • Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).

Handling length 2^31 and more in R 3.0.0

From R News for 3.0.0 release:

There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.

In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error

> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector

However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):

> system.time(x <- seq(1,2^31))
   user  system elapsed
  8.604  11.060 120.815
> length(x)
[1] 2147483648
> length(x)/2^20
[1] 2048
> gc()
             used    (Mb) gc trigger    (Mb)   max used    (Mb)
Ncells     183823     9.9     407500    21.8     350000    18.7
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
>

Note:

  1. 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
  2. On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
  3. My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
  4. For large dataset, it makes sense to use database or specially crafted packages like bigmemory or ff or bigstatsr.
  5. [[<- for index 2^31 fails

NA in index

  • Question: what is seq(1, 3)[c(1, 2, NA)]?

Answer: It will reserve the element with NA in indexing and return the value NA for it.

  • Question: What is TRUE & NA?

Answer: NA

  • Question: What is FALSE & NA?

Answer: FALSE

  • Question: c("A", "B", NA) != "" ?

Answer: TRUE TRUE NA

  • Question: which(c("A", "B", NA) != "") ?

Answer: 1 2

  • Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?

Answer: TRUE TRUE FALSE

  • Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?

Answer: TRUE TRUE FALSE

Conclusion: In order to exclude empty or NA for numerical or character data type, we can use which() or a convenience function keep.complete(x) <- function(x) x != "" & !is.na(x). This will guarantee return logical values and not contain NAs.

Don't just use x != "" OR !is.na(x).

Some functions

Constant and 'L'

Add 'L' after a constant. For example,

for(i in 1L:n) { }

if (max.lines > 0L) { }

label <- paste0(n-i+1L, ": ")

n <- length(x);  if(n == 0L) { }

Vector/Arrays

R indexes arrays from 1 like Fortran, not from 0 like C or Python.

remove integer(0)

How to remove integer(0) from a vector?

Append some elements

append() and its after argument

setNames()

Assign names to a vector

z <- setNames(1:3, c("a", "b", "c"))
# OR
z <- 1:3; names(z) <- c("a", "b", "c")
# OR
z <- c("a"=1, "b"=2, "c"=3) # not work if "a", "b", "c" is like x[1], x[2], x[3].

Factor

labels argument

We can specify the factor levels and new labels using the factor() function.

sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))

factor(rev(letters[1:3]), labels = c("A", "B", "C"))
# C B A
# Levels: A B C

Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when()

cut(
     c(0, 10, 30), 
     breaks = c(0, 30, 50, Inf), 
     labels = c("Young", "Middle-aged", "Elderly")
 )  # Default include.lowest = FALSE
# [1] <NA>  Young Young
  • ?cut
    set.seed(1)
    x <- rnorm(100)
    facVar <- cut(x, c(min(x), -1, 1, max(x)), labels = c("low", "medium", "high"))
    table(facVar, useNA = "ifany")
    facVar
    #   low medium   high   <NA> 
    #    10     74     15      1 
    

    Note the option include.lowest = TRUE is needed when we use cut() + quantile(); otherwise the smallest data will become NA since the intervals have the format (a, b].

    x2 <- cut(x, quantile(x, 0:2/2), include.lowest = TRUE) # split x into 2 levels
    x2 <- cut(x, quantile(x, 0:3/3), include.lowest = TRUE) # split x into 3 levels
    
    library(tidyverse); library(magrittr)
    set.seed(1)
    breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
    x <- runif(50)
    bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
    
    data.frame(sc=x, bins=bins) %>% 
      group_by(bins) %>% 
      summarise(n=n()) %>% 
      ggplot(aes(x = bins, y = n)) + 
        geom_col(color = "black", fill = "#90AACB") + 
        theme_minimal() + 
        theme(axis.text.x = element_text(angle = 90)) + 
        theme(legend.position = "none") + coord_flip()
    
  • A Guide to Using the cut() Function in R
  • tibble object
    library(tidyverse)
    tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
           age_cat = case_when(
              age_yrs < 2 ~ "baby",
              age_yrs < 13 ~ "kid",
              age_yrs < 20 ~ "teen",
              TRUE         ~ "adult")
    )
    
  • R tip: Learn dplyr’s case_when() function
    case_when(
      condition1 ~ value1, 
      condition2 ~ value2,
      TRUE ~ ValueAnythingElse
    )
    # Example
    case_when(
      x %%2 == 0 ~ "even",
      x %%2 == 1 ~ "odd",
      TRUE ~ "Neither even or odd"
    )
    

How to change one of the level to NA

https://stackoverflow.com/a/25354985. Note that the factor level is removed.

x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA

Creating missing values in factors

Concatenating two factor vectors

Not trivial. How to concatenate factors, without them being converted to integer level?.

unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))

droplevels()

droplevels(): drop unused levels from a factor or, more commonly, from factors in a data frame.

factor(x , levels = ...) vs levels(x) <-

Note levels(x) is to set/rename levels, not reorder. Use relevel() or factor() to reorder.

levels()
plyr::revalue()
forcats::fct_recode()
rename levels
factor(, levels) reorder levels
sizes <- factor(c("small", "large", "large", "small", "medium"))
sizes
#> [1] small  large  large  small  medium
#> Levels: large medium small

sizes2 <- factor(sizes, levels = c("small", "medium", "large")) # reorder levels but data is not changed
sizes2
# [1] small  large  large  small  medium
# Levels: small medium large

sizes3 <- sizes
levels(sizes3) <- c("small", "medium", "large") # rename, not reorder
                                                # large -> small
                                                # medium -> medium
                                                # small -> large 
sizes3
# [1] large  small  small  large  medium
# Levels: small medium large

A regression example.

set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
 
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x2           0.99620    0.09462   10.53   <2e-16 ***

# Wrong way when we want to change the baseline level to '2'
# No change on the model fitting except the apparent change on the variable name in the printout
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x1           0.99620    0.09462   10.53   <2e-16 ***

# Correct way if we want to change the baseline level to '2'
# The estimate was changed by flipping the sign from the original data
sample_data$x <- relevel(x, ref = "2")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  1.96425    0.06770   29.01   <2e-16 ***
# x1          -0.99620    0.09462  -10.53   <2e-16 ***

stats::relevel()

relevel. This function can only be used to change the reference level of a factor variable. It does not directly create an arbitrary order of levels. That is, it is useful in lm() or aov(), etc.

reorder(), levels() and boxplot()

  • How to Reorder Boxplots in R: A Comprehensive Guide (tapply() method, simple & effective)
  • reorder().This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
    # Syntax:
    # newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
    
    (bymedian <- with(InsectSprays, reorder(spray, count, median)) )
    class(bymedian)
    levels(bymedian)
    boxplot(count ~ bymedian, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # boxplots are sorted according to the new levels
    boxplot(count ~ spray, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # not sorted
    
  • Statistics Sunday: My 2019 Reading (reorder function)

factor() vs ordered()

factor(levels=c("a", "b", "c"), ordered=TRUE)
# ordered(0)
# Levels: a < b < c

factor(levels=c("a", "b", "c"))
# factor(0)
# Levels: a b c

ordered(levels=c("a", "b", "c"))
# Error in factor(x, ..., ordered = TRUE) : 
#  argument "x" is missing, with no default

Data frame

stringsAsFactors = FALSE

http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/

We can use options(stringsAsFactors=FALSE) forces R to import character data as character objects.

In R 4.0.0, stringAsFactors=FALSE will be default. This also affects read.table() function.

check.names = FALSE

Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.

> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
  1a 2a
1  1  1
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1   1   1
2   2   2

Create unique rownames: make.unique()

groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
rownames(mydf) <- make.unique(groupCodes)

data.frame() will change rownames

class(df2)
# [1] "matrix" "array"
rownames(df2)[c(9109, 44999)]
# [1] "A1CF"     "A1BG-AS1"
rownames(data.frame(df2))[c(9109, 44999)]
# [1] "A1CF"     "A1BG.AS1"

Print a data frame without rownames

# Method 1. 
rownames(df1) <- NULL

# Method 2. 
print(df1, row.names = FALSE)

Convert data frame factor columns to characters

Convert data.frame columns from factors to characters

# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)

# Method 2:
bob[] <- lapply(bob, as.character)

To replace only factor columns:

# Method 1:
i <- sapply(bob, is.factor)
bob[i] <- lapply(bob[i], as.character)

# Method 2:
library(dplyr)
bob %>% mutate_if(is.factor, as.character) -> bob

Sort Or Order A Data Frame

How To Sort Or Order A Data Frame In R

  1. df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
  2. library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
  3. library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
  4. library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)

data.frame to vector

df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))

class(df)
# [1] "data.frame"
class(t(df))
# [1] "matrix" "array"
class(unlist(df))
# [1] "numeric"

# Method 1: Convert data frame to matrix using as.matrix()
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)   # [1] 1 2 3 4 5 6
vec2 <- c(mat)

# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))

# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6

# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3 
#  1  2  3  4  5  6 
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6

Q: Why as.vector(df) cannot convert a data frame into a vector?

A: The as.vector function cannot be used directly on a data frame to convert it into a vector because a data frame is a list of vectors (i.e., its columns) and as.vector only removes the attributes of an object to create a vector. When you apply as.vector to a data frame, R does not know how to concatenate these independent columns (which could be of different types) into a single vector. Therefore, it doesn’t perform the operation. Therefore as.vector() returns the underlying list structure of the data frame instead of converting it into a vector.

However, when you transpose the data frame using t(), it gets converted into a matrix. A matrix in R is a vector with dimensions. Therefore, all elements of the matrix must be of the same type. If they are not, R will coerce them to be so. Once you have a matrix, as.vector() can easily convert it into a vector because all elements are of the same type.

Using cbind() to merge vectors together?

It’s a common mistake to try and create a data frame by cbind()ing vectors together. This doesn’t work because cbind() will create a matrix unless one of the arguments is already a data frame. Instead use data.frame() directly. See Advanced R -> Data structures chapter.

cbind NULL and data.frame

cbind can't combine NULL with dataframe. Add as.matrix() will fix the problem.

merge

Special character in the matched variable can create a trouble when we use merge() or dplyr::inner_join(). I guess R internally turns df2 (a matrix but not a data frame) to a data frame (so rownames are changed if they contain special character like "-"). This still does not explain the situation when I

class(df1); class(df2)
# [1] "data.frame"  # 2 x 2
# [1] "matrix" "array" # 52439 x 2
rownames(df1)
# [1] "A1CF"     "A1BG-AS1"
merge(df1, df2[c(9109, 44999), ], by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BG-AS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1      A1CF     0    0    3.01147   1.189578
rownames(df2)[38958:38959]
# [1] "ITFG2-AS1"  "ADGRD1-AS1"

rownames(df1)[2] <- "A1BGAS1"
rownames(df2)[44999] <- "A1BGAS1"
merge(df1, df2, by= 0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1   A1BGAS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578

is.matrix: data.frame is not necessarily a matrix

See ?matrix. is.matrix returns TRUE if x is a vector and has a "dim" attribute of length 2 and FALSE otherwise.

An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by

X <- data.frame(x=1:2, y=3:4)

The 'X' object is NOT a vector and it does NOT have the "dim" attribute. It has only 3 attributes: "names", "row.names" & "class". Note that dim() function works fine and returns correctly though there is not "dim" attribute.

Another example that is a data frame but not a matrix is the built-in object cars; see ?matrix. It is not a vector

Convert a data frame to a matrix: as.matrix() vs data.matrix()

If I have a data frame X which recorded the time of some files.

  • is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
  • as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
  • data.matrix(X) will convert the time to numerical values. So use data.matrix() if the data is numeric. The returned object is not a data frame anymore.
# latex directory contains cache files from knitting an rmarkdown file
X <- list.files("latex/", full.names = T) %>%
     grep("RData", ., value=T) %>% 
     file.info() %>%  
     `[`("mtime")
X %>% is.data.frame() # TRUE
X %>% is.matrix() # FALSE
X %>% as.matrix() %>% is.matrix() # TRUE
X %>% data.matrix() %>% is.matrix() # TRUE
X %>% as.matrix() %>% "["(1:2, ) # timestamps
X %>% data.matrix() %>% "["(1:2, ) # numeric
  • The as.matrix() function is used to coerce an object into a matrix. It can be used with various types of R objects, such as vectors, data frames, and arrays.
  • The data.matrix() function is specifically designed for converting a data frame into a matrix by coercing all columns to numeric values. If the data frame contains non-numeric columns, such as character or factor columns, data.matrix() will convert them to numeric values if possible (e.g., by converting factors to their integer codes).
  • See the following example where as.matrix() and data.matrix() return different resuls.
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
mat <- as.matrix(df)
mat
#      a   b  
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array" 
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array" 
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"

matrix vs data.frame

Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.

Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
 [1] "time"   "status" "treat"  "AKT1"   "BRAF"   "FLOT2"  "MTOR"   "PCK2"   "PIK3CA"
[10] "RAF1"  
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10

Case 2:

ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])   # OK

ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
unique(ip2$Priority)     # OK

The length of a matrix and a data frame is different.

> length(matrix(1:6, 3, 2))
[1] 6
> length(data.frame(matrix(1:6, 3, 2)))
[1] 2
> x[1]
  X1
1  1
2  2
3  3
4  4
5  5
6  6
> x1
[1] 1 2 3 4 5 6

So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.

How to Remove Duplicates

How to Remove Duplicates in R with Example

Convert a matrix (not data frame) of characters to numeric

Just change the mode of the object

tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
     a     b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)     # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"

> mode(tmp) <- "numeric"
> sum(tmp)
[1] 1.917

Convert Data Frame Row to Vector

as.numeric() or c()

Convert characters to integers

mode(x) <- "integer"

Non-Standard Evaluation

Understanding Non-Standard Evaluation. Part 1: The Basics

Select Data Frame Columns in R

This is part of series of DATA MANIPULATION IN R from datanovia.com

  • pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
  • select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
  • select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
  • Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names

Another way is to the dollar sign $ operator (?"$") to extract rows or column from a data frame.

class(USArrests)  # "data.frame"
USArrests$"Assault"

Note that for both data frame and matrix objects, we need to use the [ operator to extract columns and/or rows.

USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#         Murder Assault
# Alabama   13.2     236
# Alaska    10.0     263
USArrests[c("Murder", "Assault")]  # all rows

tmp <- data(package="datasets")
class(tmp$results)  # "matrix" "array" 
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix

Note for a data.table object, we can extract columns using the column names without double quotes.

data.table(USArrests)[1:2, list(Murder, Assault)]

Add columns to a data frame

How to add columns to a data frame in R

Exclude/drop/remove data frame columns

# method 1
df = subset(mydata, select = -c(x,z) )

# method 2
drop <- c("x","z")
df = mydata[,!(names(mydata) %in% drop)]

# method 3: dplyr
mydata2 = select(mydata, -a, -x, -y)
mydata2 = select(mydata, -c(a, x, y))
mydata2 = select(mydata, -a:-y)
mydata2 = mydata[,!grepl("^INC",names(mydata))]

Remove Rows from the data frame

Remove Rows from the data frame in R

Danger of selecting rows from a data frame

> dim(cars)
[1] 50  2
> data.frame(a=cars[1,], b=cars[2, ])
  a.speed a.dist b.speed b.dist
1       4      2       4     10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a  b
speed 4  4
dist  2 10

Creating data frame using structure() function

Creating data frame using structure() function in R

Create an empty data.frame

https://stackoverflow.com/questions/10689055/create-an-empty-data-frame

# the column types default as logical per vector(), but are then overridden
a = data.frame(matrix(vector(), 5, 3,
               dimnames=list(c(), c("Date", "File", "User"))),
               stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)

# similar to above
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))

# different data type
a <- data.frame(x1 = character(),
                x2 = numeric(),
                x3 = factor(),
                stringsAsFactors = FALSE)

Objects from subsetting a row in a data frame vs matrix

  • Subsetting creates repeated rows. This will create unexpected rownames.
    R> z <- data.frame(x=1:3, y=2:4)
    R> rownames(z) <- letters[1:3]
    R> rownames(z)[c(1,1)]
    [1] "a" "a"
    R> rownames(z[c(1,1),])
    [1] "a"   "a.1"
    R> z[c(1,1), ]
        x y
    a   1 2
    a.1 1 2
    
  • Convert a dataframe to a vector (by rows) The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
    str(trainData)
    # 'data.frame':	503 obs. of  500 variables:
    #  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
    #  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
    # ...
    trainData[1:3, 1:3]
    #        bm001      bm002    bm003
    # 1  0.4289449 0.05676296 1.657966
    # 2  1.0000000 1.00000000 1.000000
    # 3 -0.5000000 0.50000000 0.500000
    o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
    # Warning message:
    # In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
    #   row names were found from a short variable and have been discarded
    

    'trees' data from the 'datasets' package

    trees[1:3,]
    #   Girth Height Volume
    # 1   8.3     70   10.3
    # 2   8.6     65   10.3
    # 3   8.8     63   10.2
    
    # Wrong ways:
    data.frame(trees[1,] , trees[2,])
    #   Girth Height Volume Girth.1 Height.1 Volume.1
    # 1   8.3     70   10.3     8.6       65     10.3
    data.frame(time=trees[1,] , status=trees[2,])
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=as.vector(trees[1,]) , status=as.vector(trees[2,]))
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=c(trees[1,]) , status=c(trees[2,]))
    # time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    
    # Right ways:
    # method 1: dropping row names
    data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,]))) 
    # OR
    data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
    #   time status
    # 1  8.3    8.6
    # 2 70.0   65.0
    # 3 10.3   10.3
    # method 2: keeping row names
    data.frame(time=t(trees[1,]) , status=t(trees[2,]))
    #          X1   X2
    # Girth   8.3  8.6
    # Height 70.0 65.0
    # Volume 10.3 10.3
    data.frame(time=unlist(trees[1,]) , status=unlist(trees[2,]))
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    # Method 3: convert a data frame to a matrix
    is.matrix(trees)
    # [1] FALSE
    trees2 <- as.matrix(trees)
    data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    dim(trees[1,])
    # [1] 1 3
    dim(trees2[1, ])
    # NULL
    trees[1, ]  # notice the row name '1' on the left hand side
    #   Girth Height Volume
    # 1   8.3     70   10.3
    trees2[1, ]
    #  Girth Height Volume
    #    8.3   70.0   10.3
    

Convert a list to data frame

How to Convert a List to a Data Frame in R.

# method 1
data.frame(t(sapply(my_list,c)))

# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)

# method 3
library(data.table)
rbindlist(my_list)

tibble and data.table

Clean a dataset

How to clean the datasets in R

matrix

Define and subset a matrix

  • Matrix in R
    • It is clear when a vector becomes a matrix the data is transformed column-wisely (byrow = FALSE, by default).
    • When subsetting a matrix, it follows the format: X[rows, colums] or X[y-axis, x-axis].
data <- c(2, 4, 7, 5, 10, 1)
A <- matrix(data, ncol = 3)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7   10
# [2,]    4    5    1

A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]    7   10

Prevent automatic conversion of single column to vector

use drop = FALSE such as mat[, 1, drop = FALSE].

complete.cases(): remove rows with missing in any column

It works on a sequence of vectors, matrices and data frames.

NROW vs nrow

?nrow. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.

matrix (column-major order) multiply a vector

> matrix(1:6, 3,2)
     [,1] [,2]
[1,]    1    4
[2,]    2    5
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3) # c(1,2,3) will be recycled to form a matrix. Good quiz.
     [,1] [,2]
[1,]    1    4
[2,]    4   10
[3,]    9   18
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
     [,1] [,2]
[1,]    1   16
[2,]    4    5
[3,]    9   12

add a vector to all rows of a matrix

add a vector to all rows of a matrix. sweep() or rep() is the best.

sparse matrix

R convert matrix or data frame to sparseMatrix

To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, as.vector().

Attributes

Names

Useful functions for dealing with object names. (Un)Setting object names: stats::setNames(), unname() and rlang::set_names()

Print a vector by suppressing names

Use unname. sapply(, , USE.NAMES = FALSE).

format.pval/print p-values/format p values

format.pval(). By default it will show 5 significant digits (getOption("digits")-2).

> format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"     
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"

R> pvalue
[1] 0.0004632104
R> print(pvalue, digits =20)
[1] 0.00046321036188223807528
R> format.pval(pvalue)
[1] "0.00046321"
R> format.pval(pvalue * 1e-1)
[1] "4.6321e-05"
R> format.pval(0.00004632)
[1] "4.632e-05"
R> getOption("digits")
[1] 7

Customize R: options()

Change the default R repository, my .Rprofile

Change R repository

Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local .Rprofile settings take precedence.

For example, I can specify the R mirror I like by creating a single line .Rprofile file under my home directory. Another good choice of repository is cloud.r-project.org.

Type file.edit("~/.Rprofile")

local({
  r = getOption("repos")
  r["CRAN"] = "https://cran.rstudio.com/"
  options(repos = r)
})
options(continue = "  ", editor = "nano")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
  .Last <- function() try(savehistory("~/.Rhistory"))
}

Change the default web browser for utils::browseURL()

When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put

options(browser='seamonkey')

in the .Rprofile of your home directory. If the browser is not in the global PATH, we need to put the full path above.

For one-time only purpose, we can use the browser option in help.start() function:

> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...

We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See

Change the default editor

On my Linux and mac, the default editor is "vi". To change it to "nano",

options(editor = "nano")

Change prompt and remove '+' sign

See https://stackoverflow.com/a/1448823.

options(prompt="R> ", continue=" ")

digits

  • signif() rounds x to n significant digits.
    R> signif(pi, 3)
    [1] 3.14
    R> signif(pi, 5)
    [1] 3.1416
    
  • The default digits 7 may be too small. For example, if a number is very large, then we may not be able to see (enough) value after the decimal point. The acceptable range is 1-22. See the following examples

In R,

> options()$digits # Default
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
[1] 100000.1
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11

In Python,

>>> 100000.07 + .04
100000.11

Disable scientific notation in printing: options(scipen)

How to Turn Off Scientific Notation in R?

This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.

> numer = 29707; denom = 93874
> c(numer/denom, numer, denom) 
[1] 3.164561e-01 2.970700e+04 9.387400e+04

# Method 1. Without changing the global option
> format(c(numer/denom, numer, denom), scientific=FALSE)
[1] "    0.3164561" "29707.0000000" "93874.0000000"

# Method 2. Change the global option
> options(scipen=999)
> numer/denom
[1] 0.3164561
> c(numer/denom, numer, denom)
[1]     0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]     0.8 29707.0 93874.0

Suppress warnings: options() and capture.output()

Use options(). If warn is negative all warnings are ignored. If warn is zero (the default) warnings are stored until the top--level function returns.

op <- options("warn")
options(warn = -1)
....
options(op)

# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)

suppressWarnings()

suppressWarnings( foo() )

foo <- capture.output( 
 bar <- suppressWarnings( 
 {print( "hello, world" ); 
   warning("unwanted" )} ) ) 

capture.output()

str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")

Converts warnings into errors

options(warn=2)

demo() function

  • How to wait for a keypress in R? PS readline() is different from readLines().
    for(i in 1:2) { print(i); readline("Press [enter] to continue")}
    
  • Hit 'ESC' or Ctrl+c to skip the prompt "Hit <Return> to see next plot:"
  • demo() uses options() to ask users to hit Enter on each plot
    op <- options(device.ask.default = ask)  # ask = TRUE
    on.exit(options(op), add = TRUE)
    

sprintf

paste, paste0, sprintf

this post, 3 R functions that I enjoy

sep vs collapse in paste()

  • sep is used if we supply multiple input objects to paste()
  • collapse is used to make the output of length 1. It is commonly used if we have only 1 input object
R> paste("a", "A", sep=",")
[1] "a,A"
R> paste("a", "A", sep=",", collapse="-")
[1] "a,A"
R> paste(c("a", "A"), collapse="-")
[1] "a-A"

R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], collapse = "-")
[1] "a-b-c"

Format number as fixed width, with leading zeros

# sprintf()
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# formatC()
paste("name", formatC(a, width=3, flag="0"), sep="_")
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# gsub()
paste0("bm", gsub(" ", "0", format(5:15)))
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"

formatC and prettyNum (prettifying numbers)

R> (x <- 1.2345 * 10 ^ (-8:4))
 [1] 1.2345e-08 1.2345e-07 1.2345e-06 1.2345e-05 1.2345e-04 1.2345e-03
 [7] 1.2345e-02 1.2345e-01 1.2345e+00 1.2345e+01 1.2345e+02 1.2345e+03
[13] 1.2345e+04
R> formatC(x)
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "0.0001234" "0.001234"
 [7] "0.01235"   "0.1235"    "1.234"     "12.34"     "123.4"     "1234"
[13] "1.234e+04"
R> formatC(x, digits=3)
 [1] "1.23e-08" "1.23e-07" "1.23e-06" "1.23e-05" "0.000123" "0.00123"
 [7] "0.0123"   "0.123"    "1.23"     "12.3"     " 123"     "1.23e+03"
[13] "1.23e+04"
R> formatC(x, digits=3, format="e")
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "1.234e-04" "1.234e-03"
 [7] "1.235e-02" "1.235e-01" "1.234e+00" "1.234e+01" "1.234e+02" "1.234e+03"
[13] "1.234e+04"

R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"

Format(x, scientific = TRUE)

Print numeric data in exponential format, so .0001 prints as 1e-4

Creating publication quality graphs in R

HDF5 : Hierarchical Data Format

HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.

Formats for writing/saving and sharing data

Efficiently Saving and Sharing Data in R

Write unix format files on Windows and vice versa

https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html

with() and within() functions

closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)

mk <- within(mariokart, {
             closePr <- totalPr - shipPr
     })
head(mk) # new column closePr

mk <- mariokart
aggregate(. ~ wheels + cond, mk, mean)
# create mean according to each level of (wheels, cond)

aggregate(totalPr ~ wheels + cond, mk, mean)

tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)

stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals

Plot histograms as lines

https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.

x2=invisible(hist(out2$EB))
y2=invisible(hist(out2$Bench))
z2=invisible(hist(out2$EB0.001))

plot(x=x2$mids, y=x2$density, type="l")
lines(y2$mids, y2$density, lty=2, pwd=2)
lines(z2$mids, z2$density, lty=3, pwd=2)

Histogram with density line

hist(x, prob = TRUE)
lines(density(x), col = 4, lwd = 2)

The overlayed density may looks strange in cases for example counts from single-cell RNASeq or p-values from RNASeq (there is a peak around x=0).

Graphical Parameters, Axes and Text, Combining Plots

statmethods.net

15 Questions All R Users Have About Plots

See 15 Questions All R Users Have About Plots. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.

  1. How To Draw An Empty R Plot? plot.new()
  2. How To Set The Axis Labels And Title Of The R Plots?
  3. How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
  4. How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). ?par.
  5. How To Add Or Change The R Plot’s Legend? legend()
  6. How To Draw A Grid In Your R Plot? grid()
  7. How To Draw A Plot With A PNG As Background? rasterImage() from the png package
  8. How To Adjust The Size Of Points In An R Plot? cex argument
  9. How To Fit A Smooth Curve To Your R Data? loess() and lines()
  10. How To Add Error Bars In An R Plot? arrows()
  11. How To Save A Plot As An Image On Disc
  12. How To Plot Two R Plots Next To Each Other? par(mfrow)[which means Multiple Figures (use ROW-wise)], gridBase package, lattice package
  13. How To Plot Multiple Lines Or Points? plot(), lines()
  14. How To Fix The Aspect Ratio For Your R Plots? asp parameter
  15. What Is The Function Of hjust And vjust In ggplot2?

jitter function

Jitterbox.png

Scatterplot with the "rug" function

require(stats)  # both 'density' and its default method
with(faithful, {
    plot(density(eruptions, bw = 0.15))
    rug(eruptions)
    rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
})

File:RugFunction.png

See also the stripchart() function which produces one dimensional scatter plots (or dot plots) of the given data.

Identify/Locate Points in a Scatter Plot

  • ?identify
  • Using the identify function in R
    plot(x, y)
    identify(x, y, labels = names, plot = TRUE) 
    # Use left clicks to select points we want to identify and "esc" to stop the process
    # This will put the labels on the plot and also return the indices of points
    # [1] 143
    names[143]
    

Draw a single plot with two different y-axes

Draw Color Palette

Default palette before R 4.0

palette() # black, red, green3, blue, cyan, magenta, yellow, gray

# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
palette(tropical)
plot(1:5, 1:5, col=1:5, pch=16, cex=5)

New palette in R 4.0.0

R 4.0: 3 new features, R 4.0.0 now available, and a look back at R's history. For example, we can select "ggplot2" palette to make the base graphics charts that match the color scheme of ggplot2.

R> palette() 
[1] "black"   "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "gray62"
R> palette.pals()
 [1] "R3"              "R4"              "ggplot2"        
 [4] "Okabe-Ito"       "Accent"          "Dark 2"         
 [7] "Paired"          "Pastel 1"        "Pastel 2"       
[10] "Set 1"           "Set 2"           "Set 3"          
[13] "Tableau 10"      "Classic Tableau" "Polychrome 36"  
[16] "Alphabet"
R> palette.colors(palette='R4') # same as palette()
[1] "#000000" "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "#9E9E9E"
R> palette("R3")  # nothing return on screen but palette has changed
R> palette() 
[1] "black"   "red"     "green3"  "blue"    "cyan"    "magenta" "yellow" 
[8] "gray"  
R> palette("R4") # reset to the default color palette; OR palette("default")

R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
R> for(id in palette.pals()) { 
     scales::show_col(palette.colors(palette = id))
     title(id)
     readline("Press [enter] to continue") 
   } 

The palette function can also be used to change the color palette. See Setting up Color Palettes in R

palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
   # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))

cc <- palette()
palette(c(cc,"purple","brown")) # Add two colors
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502

evoPalette

Evolve new colour palettes in R with evoPalette

rtist

rtist: Use the palettes of famous artists in your own visualizations.

SVG

Embed svg in html

svglite

svglite is better R's svg(). It was used by ggsave(). svglite 1.2.0, R Graphics Cookbook.

pdf -> svg

Using Inkscape. See this post.

svg -> png

SVG to PNG using the gyro package

read.table

clipboard

source("clipboard")
read.table("clipboard")

inline text

mydf <- read.table(header=T, text='
 cond yval
    A 2
    B 2.5
    C 1.6
')

http(s) connection

temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt", 
                           ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)

read only specific columns

Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.

x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)), 
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
# 
system.time(x <- read.delim("Methylation450k.txt", 
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))

To know the number of columns, we might want to read the first row first.

library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()

Another method is to use pipe(), cut or awk. See ways to read only selected columns from a file into R

check.names = FALSE in read.table()

gx <- read.table(file, header = T, row.names =1)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH."        "IGHV1.69"    "IGKV1.5"     "IGKV2.24"    "KRTAP13.2"  
# [7] "KRTAP19.1"   "KRTAP2.4"    "KRTAP5.9"    "KRTAP6.3"    "Kua.UEV"  

gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH@"        "IGHV1-69"    "IGKV1-5"     "IGKV2-24"    "KRTAP13-2"  
# [7] "KRTAP19-1"   "KRTAP2-4"    "KRTAP5-9"    "KRTAP6-3"    "Kua-UEV"  

setNames()

Change the colnames. See an example from tidymodels

Testing for valid variable names

Testing for valid variable names

make.names(): Make syntactically valid names out of character vectors

  • make.names()
  • A valid variable name consists of letters, numbers and the dot or underline characters. The variable name starts with a letter or the dot not followed by a number. See R variables.
make.names("abc-d") # [1] "abc.d"

Serialization

If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the post on R mailing list.

> a <- list(1,2,3)
> a_serial <- serialize(a, NULL)
> a_length <- length(a_serial)
> a_length
[1] 70
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)

In C++ process, I receive one int variable first to get the length, and then read <length> bytes from the connection.

socketConnection

See ?socketconnection.

Simple example

from the socketConnection's manual.

Open one R session

con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
close(con1)

Open another R session (client)

con2 <- socketConnection(Sys.info()["nodename"], port = 22131)
# as non-blocking, may need to loop for input
readLines(con2)
while(isIncomplete(con2)) {
   Sys.sleep(1)
   z <- readLines(con2)
   if(length(z)) print(z)
}
close(con2)

Use nc in client

The client does not have to be the R. We can use telnet, nc, etc. See the post here. For example, on the client machine, we can issue

nc localhost 22131   [ENTER]

Then the client will wait and show anything written from the server machine. The connection from nc will be terminated once close(con1) is given.

If I use the command

nc -v -w 2 localhost -z 22130-22135

then the connection will be established for a short time which means the cursor on the server machine will be returned. If we issue the above nc command again on the client machine it will show the connection to the port 22131 is refused. PS. "-w" switch denotes the number of seconds of the timeout for connects and final net reads.

Some post I don't have a chance to read. http://digitheadslabnotebook.blogspot.com/2010/09/how-to-send-http-put-request-from-r.html

Use curl command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

curl --trace-ascii debugdump.txt http://localhost:8080/

Then go to the server,

while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")

close(con1) # return cursor in the client machine

Use telnet command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

sudo apt-get install telnet
telnet localhost 8080
abcdefg
hijklmn
qestst

Go to the server,

readLines(con1, 1)
readLines(con1, 1)
readLines(con1, 1)
close(con1) # return cursor in the client machine

Some tutorial about using telnet on http request. And this is a summary of using telnet.

Subsetting

Subset assignment of R Language Definition and Manipulation of functions.

The result of the command x[3:5] <- 13:15 is as if the following had been executed

`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)

Avoid Coercing Indices To Doubles

1 or 1L

Careful on NA value

See the example below. base::subset() or dplyr::filter() can remove NA subsets.

R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3   3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6

Implicit looping

set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]     # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7

modelling

update()

Extract all variable names in lm(), glm(), ...

all.vars(formula(Model)[-2])

as.formula(): use a string in formula in lm(), glm(), ...

? as.formula
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")

# Method 1. The 'Call' portion of the model is reported as “formula = f” 
# our modeling effort, 
# fully parameterized!
f <- as.formula(
  paste(outcome, 
        paste(variables, collapse = " + "), 
        sep = " ~ "))
print(f)
# mpg ~ cyl + disp + hp + carb

model <- lm(f, data = mtcars)
print(model)

# Call:
#   lm(formula = f, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Method 2. eval() + bquote() + ".()"
format(terms(model))  #  or model$terms
# [1] "mpg ~ cyl + disp + hp + carb"

# The new line of code
model <- eval(bquote(   lm(.(f), data = mtcars)   ))

print(model)
# Call:
#   lm(formula = mpg ~ cyl + disp + hp + carb, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Note if we skip ".()" operator
> eval(bquote(   lm(f, data = mtcars)   ))

Call:
lm(formula = f, data = mtcars)

Coefficients:
(Intercept)          cyl         disp           hp         carb  
  34.021595    -1.048523    -0.026906     0.009349    -0.926863 

reformulate

Simplifying Model Formulas with the R Function ‘reformulate()’

I() function

I() means isolates. See What does the capital letter "I" in R linear regression formula mean?, In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)

Aggregating results from linear model

https://stats.stackexchange.com/a/6862

Replacement function "fun(x) <- a"

What are Replacement Functions in R?

R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
     x[x > value] <- Inf
     x
 }
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]   1   3 Inf Inf

R> "cutoff<-"(x = xx, value = 65)
[1]   1   3 Inf Inf

The statement fun(x) <- a and R will read x <- "fun<-"(x,a)

S3 and S4 methods and signature

Debug an S4 function

  • showMethods('FUNCTION')
  • getMethod('FUNCTION', 'SIGNATURE')
  • debug(, signature)
> args(debug)
function (fun, text = "", condition = NULL, signature = NULL) 

> library(genefilter) # Bioconductor
> showMethods("nsFilter")
Function: nsFilter (package genefilter)
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")

library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code

See the source code of normalizationFactors<- (setReplaceMethod() is used) and the source code of estimateSizeFactors(). We can see how avgTxLength was used in estimateNormFactors().

Another example

library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)

showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"

debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz, 
#      normalization = ssgsea.norm, verbose = verbose, 
#      BPPARAM = BPPARAM))

isdebugged("gsva")
# [1] TRUE
undebug(gsva)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir

class(ir)
## [1] "IRanges"
## attr(,"package")
## [1] "IRanges"

getClassDef(class(ir))
## Class "IRanges" [package "IRanges"]
## 
## Slots:
##                                                                       
## Name:            start           width           NAMES     elementType
## Class:         integer         integer characterORNULL       character
##                                       
## Name:  elementMetadata        metadata
## Class: DataTableORNULL            list
## 
## Extends: 
## Class "Ranges", directly
## Class "IntegerList", by class "Ranges", distance 2
## Class "RangesORmissing", by class "Ranges", distance 2
## Class "AtomicList", by class "Ranges", distance 3
## Class "List", by class "Ranges", distance 4
## Class "Vector", by class "Ranges", distance 5
## Class "Annotated", by class "Ranges", distance 6
## 
## Known Subclasses: "NormalIRanges"

Check if a function is an S4 method

isS4(foo)

How to access the slots of an S4 object

  • @ will let you access the slots of an S4 object.
  • Note that often the best way to do this is to not access the slot directly but rather through an accessor function (e.g. coefs() rather than digging out the coefficients with $ or @). However, often such functions do not exist so you have to access the slots directly. This will mean that your code breaks if the internal implementation changes, however.
  • R - S4 Classes and Methods Hansen. getClass() or getClassDef().

setReplaceMethod()

See what methods work on an object

see what methods work on an object, e.g. a GRanges object:

methods(class="GRanges")

Or if you have an object, x:

methods(class=class(x))

View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere()

?":::"

  • pkg::name returns the value of the exported variable name in namespace pkg
  • pkg:::name returns the value of the internal variable name
base::"+"
stats:::coef.default

predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR  
getS3method("predict", "ppr")

getS3method("t", "test")

methods() + getAnywhere() functions

Read the source code (include Fortran/C, S3 and S4 methods)

S3 method is overwritten

For example, the select() method from dplyr is overwritten by grpreg package.

An easy solution is to load grpreg before loading dplyr.

mcols() and DataFrame() from Bioc S4Vectors package

  • mcols: Get or set the metadata columns.
  • colData: SummarizedExperiment instances from GenomicRanges
  • DataFrame: The DataFrame class extends the DataTable virtual class and supports the storage of any type of object (with length and [ methods) as columns.

For example, in Shrinkage of logarithmic fold changes vignette of the DESeq2paper package

> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
   baseMean   baseVar   allZero dispGeneEst    dispFit dispersion  dispIter dispOutlier   dispMAP
  <numeric> <numeric> <logical>   <numeric>  <numeric>  <numeric> <numeric>   <logical> <numeric>
1  163.5750  8904.607     FALSE  0.06263141 0.03862798  0.0577712         7       FALSE 0.0577712
2  175.3883 59643.515     FALSE  2.25306109 0.03807917  2.2530611        12        TRUE 1.6011440
  Intercept strain_DBA.2J_vs_C57BL.6J SE_Intercept SE_strain_DBA.2J_vs_C57BL.6J WaldStatistic_Intercept
  <numeric>                 <numeric>    <numeric>                    <numeric>               <numeric>
1  6.210188                  1.735829    0.1229354                    0.1636645               50.515872
2  6.234880                  1.823173    0.6870629                    0.9481865                9.074686
  WaldStatistic_strain_DBA.2J_vs_C57BL.6J WaldPvalue_Intercept WaldPvalue_strain_DBA.2J_vs_C57BL.6J
                                <numeric>            <numeric>                            <numeric>
1                                10.60602         0.000000e+00                         2.793908e-26
2                                 1.92280         1.140054e-19                         5.450522e-02
   betaConv  betaIter  deviance  maxCooks
  <logical> <numeric> <numeric> <numeric>
1      TRUE         3  210.4045 0.2648753
2      TRUE         9  243.7455 0.3248949

Pipe

Packages take advantage of pipes

  • rstatix: Pipe-Friendly Framework for Basic Statistical Tests

findInterval()

Related functions are cuts() and split(). See also

Assign operator

  • Earlier versions of R used underscore (_) as an assignment operator.
  • Assignments with the = Operator
  • In R 1.8.0 (2003), the assign operator has been removed. See NEWS.
  • In R 1.9.0 (2004), "_" is allowed in valid names. See NEWS.
R162.png

Operator precedence

The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.

order(), rank() and sort()

If we want to find the indices of the first 25 genes with the smallest p-values, we can use order(pval)[1:25].

> x = sample(10)
> x
 [1]  4  3 10  7  5  8  6  1  9  2
> order(x)
 [1]  8 10  2  1  5  7  4  6  9  3
> rank(x)
 [1]  4  3 10  7  5  8  6  1  9  2
> rank(10*x)
 [1]  4  3 10  7  5  8  6  1  9  2

> x[order(x)]
 [1]  1  2  3  4  5  6  7  8  9 10
> sort(x)
 [1]  1  2  3  4  5  6  7  8  9 10

OS-dependent results on sorting string vector

Gene symbol case.

# mac: 
order(c("DC-UbP", "DC2")) # c(1,2)

# linux: 
order(c("DC-UbP", "DC2")) # c(2,1)

Affymetric id case.

# mac:
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"

# linux
order(c("202800_at", "2028_s_at")) # [1] 1 2
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"

It does not matter if we include factor() on the character vector.

The difference is related to locale. See

# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2

# Or setting the locale to "C"
Sys.setlocale("LC_ALL", "C"); sort(c("DC-UbP", "DC2"))
# Or
Sys.setlocale("LC_COLLATE", "C"); sort(c("DC-UbP", "DC2"))
# But not
Sys.setlocale("LC_ALL", "en_US.UTF-8"); sort(c("DC-UbP", "DC2"))

unique()

It seems it does not sort. ?unique.

# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"

do.call

do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it.

The do.call() function in R: Unlocking Efficiency and Flexibility

Below are some examples from the help.

  • Usage
do.call(what, args, quote = FALSE, envir = parent.frame())
# what: either a function or a non-empty character string naming the function to be called.
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# quote: a logical value indicating whether to quote the arguments.
# envir: an environment within which to evaluate the call. This will be most useful
#        if what is a character string and the arguments are symbols or quoted expressions.
  • do.call() is similar to lapply() but not the same. It seems do.call() can make a simple function vectorized.
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
> lapply(list(imag = 1:3), complex)
$imag
[1] 0+0i
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
> do.call(function(x) x+1, list(1:3))
[1] 2 3 4
  • Applying do.call with Multiple Arguments
> do.call("sum", list(c(1,2,3,NA), na.rm = TRUE))
[1] 6
> do.call("sum", list(c(1,2,3,NA) ))
[1] NA
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> length(tmp)
[1] 3
> tmp[1:4,]
  Var1 Var2 Var3
1    a    1    +
2    b    1    +
3    a    2    +
4    b    2    +
> c(tmp, sep = "")
$Var1
 [1] a b a b a b a b a b a b
Levels: a b

$Var2
 [1] 1 1 2 2 3 3 1 1 2 2 3 3

$Var3
 [1] + + + + + + - - - - - -
Levels: + -

$sep
[1] ""
> do.call("paste", c(tmp, sep = ""))
 [1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[12] "b3-"
  • environment and quote arguments.
> A <- 2
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A)   
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env)  
[1] 4
> do.call(f, list(A), envir = env)   
[1] 2                       # Why?

> eval(call("f", A))                      
[1] 2
> eval(call("f", quote(A)))               
[1] 2
> eval(call("f", A), envir = env)         
[1] 4
> eval(call("f", quote(A)), envir = env)  
[1] 100
> foo <- function(a=1, b=2, ...) { 
         list(arg=do.call(c, as.list(match.call())[-1])) 
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a 
1 
> foo(a=1, b=2, c=3)
$arg
a b c 
1 2 3 
  • do.call() + switch(). See an example from Seurat::NormalizeData.
do.call(
   what = switch(
     EXPR = margin,
     '1' = 'rbind',
     '2' = 'cbind',
     stop("'margin' must be 1 or 2")
   ),
   args = normalized.data
)
switch('a', 'a' = rnorm(3), 'b'=rnorm(4)) # switch returns a value
do.call(switch('a', 'a' = 'rnorm', 'b'='rexp'), args=list(n=4)) # switch returns a function
  • The function we want to call is a string that may change: glmnet
# Suppose we want to call cv.glmnet or cv.coxnet or cv.lognet or cv.elnet .... depending on the case
fun = paste("cv", subclass, sep = ".")
cvstuff = do.call(fun, list(predmat,y,type.measure,weights,foldid,grouped))

expand.grid, mapply, vapply

A faster way to generate combinations for mapply and vapply

do.call vs mapply

  • do.call() is doing what mapply() does but do.call() uses a list instead of multiple arguments. So do.call() more close to base::Map() function.
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
      Var1 
 [1,] "a1+"
 [2,] "b1+"
 [3,] "a2+"
 [4,] "b2+"
 [5,] "a3+"
 [6,] "b3+"
 [7,] "a1-"
 [8,] "b1-"
 [9,] "a2-"
[10,] "b2-"
[11,] "a3-"
[12,] "b3-"
# It does not work if we do not explicitly specify the arguments in mapply()
> mapply(paste, tmp, sep = "")
      Var1 Var2 Var3
 [1,] "a"  "1"  "+" 
 [2,] "b"  "1"  "+" 
 [3,] "a"  "2"  "+" 
 [4,] "b"  "2"  "+" 
 [5,] "a"  "3"  "+" 
 [6,] "b"  "3"  "+" 
 [7,] "a"  "1"  "-" 
 [8,] "b"  "1"  "-" 
 [9,] "a"  "2"  "-" 
[10,] "b"  "2"  "-" 
[11,] "a"  "3"  "-" 
[12,] "b"  "3"  "-" 
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
# [1] 1.326108 9.885284
set.seed(1)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
dim(x) # [1]  2 1000
rowMeans(x)
# [1]  1.032209 10.104131
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))

do.call vs lapply

What's the difference between lapply and do.call? It seems to me the best usage is combining both functions: do.call(..., lapply())

  • lapply returns a list of the same length as X, each element of which is the result of applying FUN to the corresponding element of X.
  • do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it. It is widely used, for example, to assemble lists into simpler structures (often with rbind or cbind).
  • Map applies a function to the corresponding elements of given vectors... Map is a simple wrapper to mapply which does not attempt to simplify the result, similar to Common Lisp's mapcar (with arguments being recycled, however). Future versions may allow some control of the result type.
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
[1] "numeric"

$Sepal.Width
[1] "numeric"

$Petal.Length
[1] "numeric"

$Petal.Width
[1] "numeric"

$Species
[1] "factor"

> x <- lapply(iris, class)
> do.call(c, x)
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species 
   "numeric"    "numeric"    "numeric"    "numeric"     "factor" 

https://stackoverflow.com/a/10801902

  • lapply applies a function over a list. So there will be several function calls.
  • do.call calls a function with a list of arguments (... argument) such as c() or rbind()/cbind() or sum or order or "[" or paste. So there is only one function call.
> X <- list(1:3,4:6,7:9)
> lapply(X,mean)
1
[1] 2

2
[1] 5

3
[1] 8
> do.call(sum, X)
[1] 45
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
[1] 45
> do.call(mean, X) # Error
> do.call(rbind,X)
     [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9
> lapply(X,rbind)
1
     [,1] [,2] [,3]
[1,]    1    2    3

2
     [,1] [,2] [,3]
[1,]    4    5    6

3
     [,1] [,2] [,3]
[1,]    7    8    9
> mapply(mean, X, trim=c(0,0.5,0.1))
[1] 2 5 8
> mapply(mean, X) 
[1] 2 5 8

Below is a good example to show the difference of lapply() and do.call() - Generating Random Strings.

> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0)
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0, collapse= "")
1
[1] "YDGA"

2
[1] "BWKN"

> do.call(paste0, x)
[1] "YB" "DW" "GK" "AN"

do.call + rbind + lapply

Lots of examples. See for example this one for creating a data frame from a vector.

x <- readLines(textConnection("---CLUSTER 1 ---
 3
 4
 5
 6
 ---CLUSTER 2 ---
 9
 10
 8
 11"))

 # create a list of where the 'clusters' are
 clust <- c(grep("CLUSTER", x), length(x) + 1L)

 # get size of each cluster
 clustSize <- diff(clust) - 1L

 # get cluster number
 clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])

 result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
     cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
         , Cluster = .cl
         )
     }))

 result

     Object Cluster
[1,] "3"    "1"
[2,] "4"    "1"
[3,] "5"    "1"
[4,] "6"    "1"
[5,] "9"    "2"
[6,] "10"   "2"
[7,] "8"    "2"
[8,] "11"   "2"

A 2nd example is to sort a data frame by using do.call(order, list()).

Another example is to reproduce aggregate(). aggregate() = do.call() + by().

attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# the above approach give the same result as the following
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)

Run examples

When we call help(FUN), it shows the document in the browser. The browser will show

example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed

How to get examples from help file, example()

Code examples in the R package manuals:

# How to run all examples from a man page
example(within)

# How to check your examples?
devtools::run_examples() 
testthat::test_examples()

See this post. Method 1:

example(acf, give.lines=TRUE)

Method 2:

Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)

"[" and "[[" with the sapply() function

Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.

sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)

is the same as

sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])

Dealing with dates

  • Find difference
# Convert the dates to Date objects
date1 <- as.Date("6/29/21", format="%m/%d/%y")
date2 <- as.Date("11/9/21", format="%m/%d/%y")

# Calculate the difference in days
diff_days <- as.numeric(difftime(date2, date1, units="days")) # 133
# In months
diff_days / (365.25/12)  # 4.36961   

# OR using the lubridate package
library(lubridate)
# Convert the dates to Date objects
date1 <- mdy("6/29/21")
date2 <- mdy("11/9/21")
interval(date1, date2) %/% months(1)
d1 = date()
class(d1) # "character"
d2 = Sys.Date()
class(d2) # "Date"

format(d2, "%a %b %d")

library(lubridate); ymd("20140108") # "2014-01-08 UTC"
mdy("08/04/2013") # "2013-08-04 UTC"
dmy("03-04-2013") # "2013-04-03 UTC"
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland") 
# "2011-08-03 10:15:03 NZST"
?Sys.timezone
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues

Nonstandard/non-standard evaluation, deparse/substitute and scoping

f <- function(x) {
  substitute(x)
}
f(1:10)
# 1:10
class(f(1:10)) # or mode()
# [1] "call"
g <- function(x) deparse(substitute(x))
g(1:10)
# [1] "1:10"
class(g(1:10)) # or mode()
# [1] "character"
  • quote(expr) - similar to substitute() but do nothing?? noquote - print character strings without quotes
mode(quote(1:10))
# [1] "call"
  • eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))

subset1 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR

subset2 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌 
subset2(sample_df, a == condition) # 👍
  • deparse(expr) - turns unevaluated expressions into character strings. For example,
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", " 
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                    
[4] "NULL"     

> deparse(args(lm), width=20)
[1] "function (formula, data, "        "    subset, weights, "           
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, "   
[5] "    y = FALSE, qr = TRUE, "       "    singular.ok = TRUE, "        
[7] "    contrasts = NULL, "           "    offset, ...) "               
[9] "NULL"

Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).

f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3

f1(1:3)
f2(1:3)
f3(1:3)

# Or
myfun <- function(f, a) {
    eval(parse(text = f))(a)
}
myfun("f1", 1:3)
myfun("f2", 1:3)
myfun("f3", 1:3)

# Or with lapply
method <- c("f1", "f2", "f3")
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method

library() accept both quoted and unquoted strings

How can library() accept both quoted and unquoted strings. The key lines are

  if (!character.only) 
     package <- as.character(substitute(package))

Lexical scoping

The ‘…’ argument

Functions

Function argument

Argument matching from R Language Definition manual.

Argument matching is augmented by the functions

Access to the partial matching algorithm used by R is via pmatch.

Check function arguments

Checking the inputs of your R functions: match.arg() , stopifnot()

stopifnot(): function argument sanity check

  • stopifnot(). stopifnot is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.
    stopifnot(condition1, condition2, ...)
    
  • Mining R 4.0.0 Changelog for Nuggets of Gold

Lazy evaluation in R functions arguments

R function arguments are lazy — they’re only evaluated if they’re actually used.

  • Example 1. By default, R function arguments are lazy.
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
  • Example 2. If you want to ensure that an argument is evaluated you can use force().
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
  • Example 3. Default arguments are evaluated inside the function.
f <- function(x = ls()) {
  a <- 1
  x
}

# ls() evaluated inside f:
f()
# [1] "a" "x"

# ls() evaluated in global environment:
f(ls())
# [1] "add"    "adders" "f" 
  • Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
x <- NULL
if (!is.null(x) && x > 0) {

}

Use of functions as arguments

Just Quickly: The unexpected use of functions as arguments

body()

Remove top axis title base plot

Return functions in R

anonymous function

In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a lambda function is defined without a name, while a regular function is defined with a name.

  • See Tidyverse page
  • But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, lapply(list(1,2,3), function(x) { x * x })
  • you can use lambda functions with many other functions in R that take a function as an argument. Some examples include sapply, apply, vapply, mapply, Map, Reduce, Filter, and Find. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
    Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
    
  • purrr anonymous function
  • The new pipe and anonymous function syntax in R 4.1.0
  • Functional programming from Advanced R
  • What are anonymous functions in R.
    > (function(x) x * x)(3)
    [1] 9
    > (\(x) x * x)(3)
    [1] 9

Backtick sign, infix/prefix/postfix operators

The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in Advanced R and What do backticks do in R?.

iris %>%  `[[`("Species")

infix operator.

1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix

Use with functions like sapply, e.g. sapply(1:5, `+`, 3) .

Error handling and exceptions, tryCatch(), stop(), warning() and message()

  • http://adv-r.had.co.nz/Exceptions-Debugging.html
  • Catch Me If You Can: Exception Handling in R
  • Temporarily disable warning messages
    # Method1: 
    suppressWarnings(expr)
    
    # Method 2:
    <pre>
    defaultW <- getOption("warn") 
    options(warn = -1) 
    [YOUR CODE] 
    options(warn = defaultW)
    
  • try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
    out <- try({
      a <- 1
      b <- "x"
      a + b
    })
    
    elements <- list(1:10, c(-1, 10), c(T, F), letters)
    results <- lapply(elements, log)
    is.error <- function(x) inherits(x, "try-error")
    succeeded <- !sapply(results, is.error)
    
  • tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
    tryCatch(expr, ..., finally)
    
    show_condition <- function(code) {
      tryCatch(code,
        error = function(c) "error",
        warning = function(c) "warning",
        message = function(c) "message"
      )
    }
    show_condition(stop("!"))
    #> [1] "error"
    show_condition(warning("?!"))
    #> [1] "warning"
    show_condition(message("?"))
    #> [1] "message"
    show_condition(10)
    #> [1] 10
    

    Below is another snippet from available.packages() function,

    z <- tryCatch(download.file(....), error = identity)
    if (!inherits(z, "error")) STATEMENTS
    
  • The return class from tryCatch() may not be fixed.
    result <- tryCatch({
      # Code that might generate an error or warning
      log(99)
    }, warning = function(w) {
      # Code to handle warnings
      print(paste("Warning:", w))
    }, error = function(e) {
      # Code to handle errors
      print(paste("Error:", e))
    }, finally = {
      # Code to always run, regardless of whether an error or warning occurred
      print("Finished")
    })   
    # character type. But if we remove 'finally', it will be numeric.
    
  • Capture message, warnings and errors from a R function

suppressMessages()

suppressMessages(expression)

List data type

Create an empty list

out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)

outlist <- as.list(seq(nfolds))

Using $ in R on a List

How to Use Dollar sign in R

Calling a function given a list of arguments

> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5

Descend recursively through lists

x[[c(5,3)]] is the same as x[[5]][[3]]. See ?Extract.

Avoid if-else or switch

?plot.stepfun.

y0 <- c(1,2,4,3)
sfun0  <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)

tt <- seq(0, 3, by = 0.1)
op <- par(mfrow = c(2,2))
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
     main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")

for(i in 1:3)
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)

par(op)

File:StepfunExample.svg

Open a new Window device

X11() or dev.new()

par()

?par

text size (cex) and font size on main, lab & axis

Examples (default is 1 for each of them):

  • cex.main=0.9
  • cex.sub
  • cex.lab=0.8, font.lab=2 (x/y axis labels)
  • cex.axis=0.8, font.axis=2 (axis/tick text/labels)
  • col.axis="grey50"

An quick example to increase font size (cex.lab, cex.axis, cex.main) and line width (lwd) in a line plot and cex & lwd in the legend.

plot(x=x$mids, y=x$density, type="l", 
     xlab="p-value", ylab="Density", lwd=2, 
     cex.lab=1.5, cex.axis=1.5, 
     cex.main=1.5, main = "")
lines(y$mids, y$density, lty=2, pwd=2)
lines(z$mids, z$density, lty=3, pwd=2)
legend('topright',legend = c('Method A','Method B','Method C'),
       lty=c(2,1,3), lwd=c(2,2,2), cex = 1.5, xjust = 0.5, yjust = 0.5)

ggplot2 case (default font size is 11 points):

  • plot.title
  • plot.subtitle
  • axis.title.x, axis.title.y: (x/y axis labels)
  • axis.text.x & axis.text.y: (axis/tick text/labels)
ggplot(df, aes(x, y)) +
  geom_point() +
  labs(title = "Title", subtitle = "Subtitle", x = "X-axis", y = "Y-axis") +
  theme(plot.title = element_text(size = 20),
        plot.subtitle = element_text(size = 15),
        axis.title.x = element_text(size = 15),
        axis.title.y = element_text(size = 15),
        axis.text.x = element_text(size = 10),
        axis.text.y = element_text(size = 10))

Default font

layout

reset the settings

op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1) 
....
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1

mtext (margin text) vs title

mgp (axis tick label locations or axis title)

  1. The margin line (in ‘mex’ units) for the axis title, axis labels and axis line. Note that ‘mgp[1]’ affects the axis ‘title’ whereas ‘mgp[2:3]’ affect tick mark labels. The default is ‘c(3, 1, 0)’. If we like to make the axis labels closer to an axis, we can use mgp=c(1.5, .5, 0) for example.
    • the default is c(3,1,0) which specify the margin line for the axis title, axis labels and axis line.
    • the axis title is drawn in the fourth line of the margin starting from the plot region, the axis labels are drawn in the second line and the axis line itself is the first line.
  2. Setting graph margins in R using the par() function and lots of cow milk
  3. Move Axis Label Closer to Plot in Base R (2 Examples)
  4. http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the labels/axis title (i.e. xlab and ylab in plot), the second the tick-mark labels, and third the tick marks. The default is c(3, 1, 0).

move axis title closer to axis

title(ylab="Within-cluster variance", line=0, 
      cex.lab=1.2, family="Calibri Light")

pch and point shapes

File:R pch.png

See here.

  • Full circle: pch=16
  • Display all possibilities: ggpubr::show_point_shapes()

lty (line type)

File:R lty.png

Line types in R: Ultimate Guide For R Baseplot and ggplot

See here.

ggpubr::show_line_types()

las (label style)

0: The default, parallel to the axis

1: Always horizontal

2: Perpendicular to the axis

3: Always vertical

oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots

no.readonly

R语言里par(no.readonly=TURE)括号里面这个参数什么意思?, R-par()

Non-standard fonts in postscript and pdf graphics

https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41


NULL, NA, NaN, Inf

https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/

save()/load() vs saveRDS()/readRDS() vs dput()/dget() vs dump()/source()

  1. saveRDS() can only save one R object while save() does not have this constraint.
  2. saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See this post.
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)

dput: Writes an ASCII text representation of an R object. The object name is not written (unlike dump).

$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato", 
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos", 
"ast", "trig", "platelet", "protime", "stage")

> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5, 
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L, 
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")

User 'verbose = TRUE' in load()

When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.

What are RDS files anyways

Archive Existing RDS Files

==, all.equal(), identical()

  • ==: exact match
  • all.equal: compare R objects x and y testing ‘near equality’
  • identical: The safe and reliable way to test two objects for being exactly equal.
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE

Be careful about using "==" to return an index of matches in the case of data with missing values.

R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1

See also the testhat package.

I found a case when I compare two objects where 1 is generated in Linux and the other is generated in macOS that identical() gives FALSE but all.equal() returns TRUE. The difference has a magnitude only e-17.

waldo

diffobj: Compare/Diff R Objects

https://cran.r-project.org/web/packages/diffobj/index.html

testthat

tinytest

tinytest: Lightweight but Feature Complete Unit Testing Framework

ttdo adds support of the 'diffobj' package for 'diff'-style comparison of R objects.

Numerical Pitfall

Numerical pitfalls in computing variance

.1 - .3/3
## [1] 0.00000000000000001388

Sys.getpid()

This can be used to monitor R process memory usage or stop the R process. See this post.

Sys.getenv() & make the script more portable

Replace all the secrets from the script and replace them with Sys.getenv("secretname"). You can save the secrets in an .Renviron file next to the script in the same project.

$ for v in 1 2; do MY=$v Rscript -e "Sys.getenv('MY')"; done
[1] "1"
[1] "2"
$ echo $MY
2

How to write R codes

  • Code smells and feels from R Consortium
    • write simple conditions,
    • handle class properly,
    • return and exit early,
    • polymorphism,
    • switch() [e.g., switch(var, value1=out1, value2=out2, value3=out3). Several examples in glmnet ]
    • case_when(),
    • %||%.
  • 5 Tips for Writing Clean R Code – Leave Your Code Reviewer Commentless
    • Comments
    • Strings
    • Loops
    • Code Sharing
    • Good Programming Practices

How to debug an R code

Debug R

Locale bug (grep did not handle UTF-8 properly PR#16264)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264

Path length in dir.create() (PR#17206)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)

install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths()

R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"

On Mac & R 3.4.0 (it's fine)

> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"

On Linux & R 3.3.1 (ARM)

> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"

On Linux & R 3.4.1 (*Problematic*)

> Sys.getenv("R_LIBS_USER")
[1] ""
> .libPaths()
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"

I need to specify the lib parameter when I use the install.packages command.

> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
Error in library(devtools) : there is no package called 'devtools'

# Specify lib.loc parameter will not help with the dependency package
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
Error: package or namespace load failed for 'devtools':
 .onLoad failed in loadNamespace() for 'devtools', details:
  call: loadNamespace(name)
  error: there is no package called 'withr'

# A solution is to redefine .libPaths
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works

A better solution is to specify R_LIBS_USER in ~/.Renviron file or ~/.bash_profile; see ?Startup.

Using external data from within another package

https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/

How to run R scripts from the command line

https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line

How to exit a sourced R script

Decimal point & decimal comma

Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark

setting seed locally (not globally) in R

https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r

R's internal C API

https://github.com/hadley/r-internals

cleancall package for C resource cleanup

Resource Cleanup in C and the R API

Random number generator

#include <R.h>

void myunif(){
  GetRNGstate();
  double u = unif_rand();
  PutRNGstate();
  Rprintf("%f\n",u);
}
$ R CMD SHLIB r_rand.c
$ R
R> dyn.load("r_rand.so")
R> set.seed(1)
R> .C("myunif")
0.265509
list()
R> .C("myunif")
0.372124
list()
R> set.seed(1)
R> .C("myunif")
0.265509
list()

Test For Randomness

Different results in Mac and Linux

Random numbers: multivariate normal

Why MASS::mvrnorm() gives different result on Mac and Linux/Windows?

The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See

rle() running length encoding

citation()

citation()
citation("MASS")
toBibtex(citation())

R not responding request to interrupt stop process

R not responding request to interrupt stop process. R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts. It seems to match with the case I'm running (dist() function).

Monitor memory usage

  • x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
  • Windows: memory.size(max=TRUE)
  • Linux
    • RStudio: htop -p PID where PID is the process ID of /usr/lib/rstudio/bin/rsession, not /usr/lib/rstudio/bin/rstudio. This is obtained by running x <- rnorm(2*1e8). The object size can be obtained through print(object.size(x), units = "auto"). Note that 1e8*8/2^20 = 762.9395.
    • R: htop -p PID where PID is the process ID of /usr/lib/R/bin/exec/R. Alternatively, use htop -p `pgrep -f /usr/lib/R/bin/exec/R`
    • To find the peak memory usage grep VmPeak /proc/$PID/status
  • mem_used() function from pryr package. It is not correct or useful if I use it to check the value compared to the memory returned by jobload in biowulf. So I cannot use it to see the memory used in running mclapply().
  • peakRAM: Monitor the Total and Peak RAM Used by an Expression or Function
  • Error: protect () : protection stack overflow and ?Memory

References:

Monitor Data

Monitoring Data in R with the lumberjack Package

Pushover

Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}

pushoverr

Resource

Books

  • Efficient R programming by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the Appendix. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
    # R 3.4.1
    .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
    setwd("/tmp/efficientR/")
    bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
    # generated pdf file is located _book/_main.pdf
    
    bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
    # generated epub file is located _book/_main.epub.
    # This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
    # but it is OK to run in an R terminal
    

Videos

“Do More with R” video tutorials. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.

Webinar

useR!

R consortium

https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured

Blogs, Tips, Socials, Communities

Bug Tracking System

https://bugs.r-project.org/bugzilla3/ and Search existing bug reports. Remember to select 'All' in the Status drop-down list.

Use sessionInfo().

License

Some Notes on GNU Licenses in R Packages

Why Dash uses the mit license (and not a copyleft gpl license)

Interview questions

  • Does R store matrices in column-major order or row-major order?
    • Matrices are stored in column-major order, which means that elements are arranged and accessed by columns. This is in contrast to languages like Python, where matrices (or arrays) are typically stored in row-major order.
  • Explain the difference between == and === in R. Provide an example to illustrate their use.
    • The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.
  • What is the purpose of the apply() function in R? How does it differ from the for loop?
    • The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.
  • Describe the concept of factors in R. How are they used in data manipulation and analysis?
    • Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.
  • What is the significance
of the attach() and detach() functions in R? When should they be used?
    • A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.
  • Explain the concept of vectorization in R. How does it impact the performance of R code?
    • Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
  • Describe the difference between data.frame and matrix in R. When would you use one over the other?
    • A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
    • A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
    • You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.
  • What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
    • The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
    • It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
    • To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.