R: Difference between revisions

From 太極
Jump to navigation Jump to search
(760 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]]


== Rmarkdown: create HTML5 web, slides and more ==
= Parallel Computing =
[[Rmarkdown]]
See [[R_parallel|R parallel]].


== [http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol HTTP protocol] ==
= Cloud Computing =


* http://en.wikipedia.org/wiki/File:Http_request_telnet_ubuntu.png
== Install R on Amazon EC2 ==
* [http://en.wikipedia.org/wiki/Query_string Query string]
http://randyzwitch.com/r-amazon-ec2/
* 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:
== Bioconductor on Amazon EC2 ==
http://www.bioconductor.org/help/bioconductor-cloud-ami/


# Open port 80 for listening
= Big Data Analysis =
# When contact is made, gather a little information (get mainly - you can ignore the rest for now)
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
# Translate the request into a file request
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
# Open the file and spit it back at the client
* [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


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.
== bigmemory, biganalytics, bigtabulate ==


=== Example in R ===
== ff, ffbase ==
<syntaxhighlight lang='r'>
* 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]
> co <- socketConnection(port=8080, server=TRUE, blocking=TRUE)
* [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]
> # Now open a web browser and type http://localhost:8080/index.html
* [http://www.bnosac.be/images/bnosac/blog/user2013_presentation_ffbase.pdf ffbase: statistical functions for large datasets] in useR 2013
> readLines(co,1)
* [https://www.rdocumentation.org/packages/ffbase/versions/0.12.7/topics/ffbase-package ffbase] package
[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) ===
== biglm ==


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/)
== data.table ==
See [[Tidyverse#data.table|data.table]].


Launch the server program (assume we have done ''gcc http_server.c -o http_server'')
== disk.frame ==
<pre>
[https://www.brodrigues.co/blog/2019-10-05-parallel_maxlik/ Split-apply-combine for Maximum Likelihood Estimation of a linear model]
$ ./http_server -p 50002
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
== Apache arrow ==
<pre>
* https://arrow.apache.org/docs/r/
GET /index.html HTTP/1.1
* [https://www.infoworld.com/article/3637038/the-best-open-source-software-of-2021.html#slide17 The best open source software of 2021]
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/index.html
= Reproducible Research =
GET /favicon.ico HTTP/1.1
* http://cran.r-project.org/web/views/ReproducibleResearch.html
Host: localhost:50002
* [[Reproducible|Reproducible]]
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
== Reproducible Environments ==
GET /favicon.ico HTTP/1.1
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
Host: localhost:50003
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
== checkpoint package ==
</pre>
* https://cran.r-project.org/web/packages/checkpoint/index.html
The browser will show the page from <index.html> in server.
* [https://timogrossenbacher.ch/2017/07/a-truly-reproducible-r-workflow/ A (truly) reproducible R workflow]


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'''.
== Some lessons in R coding ==
# 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.
# 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!


= Useful R packages =
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
* [https://github.com/qinwf/awesome-R awesome-R]
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]


=== Another Example in C (55 lines) ===
== Rcpp ==
http://mwaidyanatha.blogspot.com/2011/05/writing-simple-web-server-in-c.html  
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].


The response is embedded in the C code.  
== RInside : embed R in C++ code ==
* http://dirk.eddelbuettel.com/code/rinside.html
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
 
=== 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 [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]]


If we test the server program by opening a browser and type "http://localhost:15000/", the server received the follwing 7 lines
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>
<pre>
GET / HTTP/1.1
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
Host: localhost:15000
make
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
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>
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]).


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".
=== Windows 7 ===
 
To make RInside works on Windows OS, try the following
If we use telnet program to test, wee need to type anything we want
# 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''.
# 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 ])
# 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>
cd C:\R\R-3.0.1\library\RInside\examples\standard
make -f Makefile.win
</pre>
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
<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
* http://www.mail-archive.com/rcpp-devel@lists.r-forge.r-project.org/msg04377.html
So the Qt and Wt web tool applications on Windows may or may not be possible.


=== Others  ===
== GUI ==
* http://rosettacode.org/wiki/Hello_world/ (Different languages)
=== Qt and R ===
* http://kperisetla.blogspot.com/2012/07/simple-http-web-server-in-c.html (Windows web server)
* 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://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.)
* http://qtinterfaces.r-forge.r-project.org
* https://github.com/gtungatkar/Simple-HTTP-server
* https://github.com/davidmoreno/onion


== shiny ==
== tkrplot ==
See [[Shiny|Shiny]].
On Ubuntu, we need to install tk packages, such as by
<pre>
sudo apt-get install tk-dev
</pre>


== [https://www.rplumber.io/ plumber]: Turning your R code into a RESTful Web API ==
== reticulate - Interface to 'Python' ==
* https://github.com/trestletech/plumber
[[Python#R_and_Python:_reticulate_package|Python -> reticulate]]
* https://www.rstudio.com/resources/videos/plumber-turning-your-r-code-into-an-api/
* [https://blog.rstudio.com/2018/10/23/rstudio-1-2-preview-plumber-integration/ RStudio 1.2 Preview: Plumber Integration]
* [https://medium.com/@skyetetra/using-docker-to-deploy-an-r-plumber-api-863ccf91516d Using docker to deploy an R plumber API]


== Docker ==
== Hadoop (eg ~100 terabytes) ==
* There are two major Docker images. They include gcc, gfortran, .... So it can be used to install Rcpp package for example.
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
** [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] ==
* RHadoop
http and WebSocket library.
* 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


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.
=== [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://rapache.net/ RApache] ==
=== 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://cran.r-project.org/web/packages/gWidgetsWWW/index.html gWidgetsWWW] ==
== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
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>


* http://www.jstatsoft.org/v49/i10/paper
On CentOS,
* [https://github.com/jverzani/gWidgetsWWW2 gWidgetsWWW2] gWidgetsWWW based on Rook
<pre>
* [http://www.r-statistics.com/2012/11/comparing-shiny-with-gwidgetswww2-rapache/ Compare shiny with gWidgetsWWW2.rapache]
yum -y install libxml2 libxml2-devel
</pre>


== [http://cran.r-project.org/web/packages/Rook/index.html Rook] ==
=== XML ===
See [[Rook|Rook]].
* 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()'''.
 
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
== [https://code.google.com/p/sumo/ sumo] ==
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
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.
* https://www.tutorialspoint.com/r/r_xml_files.htm
 
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
== [http://www.stat.ucla.edu/~jeroen/stockplot Stockplot] ==
* [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)


== [http://www.rforge.net/FastRWeb/ FastRWeb] ==
# Read and parse HTML file
http://cran.r-project.org/web/packages/FastRWeb/index.html
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)


== WebDriver ==
# Extract all the paragraphs (HTML tag is p, starting at
'WebDriver' Client for 'PhantomJS'  
# the root of the document). Unlist flattens the list to
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


https://github.com/rstudio/webdriver
# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)


== [http://sysbio.mrc-bsu.cam.ac.uk/Rwui/tutorial/Instructions.html Rwui] ==
# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')
</pre>


== [http://cran.r-project.org/web/packages/CGIwithR/index.html CGHWithR] and [http://cran.r-project.org/web/packages/WebDevelopR/ WebDevelopR] ==
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.
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.
{{Pre}}
> 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"


== [http://www.rstudio.com/ide/docs/advanced/manipulate manipulate] from RStudio ==
> # try picard
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].
> 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"
</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").


Mathematica also has manipulate function for plotting; see [http://reference.wolfram.com/mathematica/tutorial/IntroductionToManipulate.html here].
=== xmlview ===
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/


== [https://github.com/att/rcloud RCloud] ==
== RCurl ==
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.
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
{{Pre}}
# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</pre>


See also the [http://user2014.stat.ucla.edu/abstracts/talks/193_Harner.pdf Talk] in UseR 2014.
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


== [https://github.com/cloudyr cloudyr] and [https://github.com/socialcopsdev/flyio flyio] - Input Output Files in R from Cloud or Local ==
No google ID is required
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 ==
Seems not work
[https://cran.r-project.org/web/packages/rdrop2/index.html rdrop2] package
<pre>
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
  arguments imply differing number of rows: 2, 0
</pre>


== Web page scraping ==
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
http://www.slideshare.net/schamber/web-data-from-r#btnNext
'''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.
{{Pre}}
# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev


=== [https://cran.r-project.org/web/packages/xml2/ xml2] package ===
# Ubuntu 16.04, 18.04
rvest package depends on xml2.
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev


=== [https://cran.r-project.org/web/packages/purrr/index.html purrr]: Functional Programming Tools ===
# Ubuntu 20.04
* https://purrr.tidyverse.org/
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev
* [http://colinfay.me/purrr-cookbook/ purrr cookbook]
</pre>
* Functional programming (cf Object-Oriented Programming)
** [http://www.youtube.com/watch?v=vLmaZxegahk Functional programming for beginners]
** [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.
* [http://staff.math.su.se/hoehle/blog/2019/01/04/mathgenius.html Purr yourself into a math genius]
* [https://martinctc.github.io/blog/vignette-write-and-read-multiple-excel-files-with-purrr/ Write & Read Multiple Excel files with purrr]


=== [https://cran.r-project.org/web/packages/rvest/index.html rvest] ===
[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://blog.rstudio.org/2014/11/24/rvest-easy-web-scraping-with-r/ Easy web scraping with R]


On Ubuntu, we need to install two packages first!
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.
<syntaxhighlight lang='bash'>
sudo apt-get install libcurl4-openssl-dev # OR libcurl4-gnutls-dev


sudo apt-get install libxml2-dev
=== [https://github.com/hadley/httr httr] ===
</syntaxhighlight>
httr imports curl, jsonlite, mime, openssl and R6 packages.


* https://github.com/hadley/rvest
When I tried to install httr package, I got an error and some message:
* [http://datascienceplus.com/visualizing-obesity-across-united-states-by-using-data-from-wikipedia/ Visualizing obesity across United States by using data from Wikipedia]
<pre>
* [https://stat4701.github.io/edav/2015/04/02/rvest_tutorial/ rvest tutorial: scraping the web using R]
Configuration failed because openssl was not found. Try installing:
* https://renkun.me/pipeR-tutorial/Examples/rvest.html
* deb: libssl-dev (Debian, Ubuntu, etc)
* http://zevross.com/blog/2015/05/19/scrape-website-data-with-the-new-r-package-rvest/
* rpm: openssl-devel (Fedora, CentOS, RHEL)
* [https://datascienceplus.com/google-scholar-scraping-with-rvest/ Google scholar scraping with rvest package]
* 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’
</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!


=== Animate ===
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://guyabel.com/post/football-kits/ Animating Changes in Football Kits using R]: rvest, tidyverse, xml2, purrr & magick
* [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 ===
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://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://cran.r-project.org/web/packages/pubmed.mineR/index.html pubmed.mineR] ===
[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)
Text mining of PubMed Abstracts (http://www.ncbi.nlm.nih.gov/pubmed). The algorithms are designed for two formats (text and XML) from PubMed.


[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)
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
curl is independent of RCurl package.


== These R packages import sports, weather, stock data and more ==
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
* https://www.computerworld.com/article/3109890/data-analytics/these-r-packages-import-sports-weather-stock-data-and-more.html
* https://www.opencpu.org/posts/curl-release-0-8/
* https://github.com/ALShum/rwunderground
* [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.
* 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 ==
{{Pre}}
https://rud.is/b/2017/02/09/diving-into-dynamic-website-content-with-splashr/
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)
</pre>


== Send email ==
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
=== [https://cran.r-project.org/web/packages/blastula/index.html blastula] ===
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
[https://rviews.rstudio.com/2019/09/04/how-to-send-custom-e-mails-with-rstudio-connect/ How to Send Custom E-mails with R]


=== [https://github.com/rpremraj/mailR/ mailR] ===
== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
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]
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).


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.
Example:
<syntaxhighlight lang='rsplus'>
{{Pre}}
> send.mail(from = "[email protected]",
# https://github.com/henrikbengtsson/matrixstats
  to = c("[email protected]", "Recipient 2 <[email protected]>"),
remotes::install_github('HenrikBengtsson/matrixStats@develop')
  replyTo = c("Reply to someone else <[email protected]>")
</pre>
  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]
== DirichletMultinomial ==
On Ubuntu, we do
<pre>
sudo apt-get install libgsl0-dev
</pre>


=== [https://cran.r-project.org/web/packages/gmailr/index.html gmailr] ===
== Create GUI ==
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://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
<syntaxhighlight lang='rsplus'>
library(gmailr)
gmail_auth('mysecret.json', scope = 'compose')


test_email <- mime() %>%
== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
  to("to@gmail.com") %>%
[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]
  from("from@gmail.com") %>%
  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] ===
== json ==
sendmailR provides a simple SMTP client. It is not clear how to use the package (i.e. where to enter the password).
[[R_web#json|R web -> json]]


=== emayili ===
== Map ==
[https://datawookie.netlify.com/blog/2019/05/emayili-sending-email-from-r/ emayili: Sending Email from R]
=== [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/


== [http://www.ncbi.nlm.nih.gov/geo/ GEO (Gene Expression Omnibus)] ==
=== choroplethr ===
See [[GEO#R_packages|this internal link]].
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
* 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/


== Interactive html output ==
=== ggplot2 ===
=== [http://cran.r-project.org/web/packages/sendplot/index.html sendplot] ===
[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/RIGHT/index.html RIGHT] ===
The supported plot types include scatterplot, barplot, box plot, line plot and pie plot.


In addition to tooltip boxes, the package can create a [http://righthelp.github.io/tutorial/interactivity table showing all information about selected nodes].
== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
See an example from [[R#RJSONIO|RJSONIO]] above.


=== r2d3 ===
== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
[https://blog.rstudio.com/2018/10/05/r2d3-r-interface-to-d3-visualizations/ r2d3] - R Interface to D3 Visualizations
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.


=== [http://cran.r-project.org/web/packages/d3Network/index.html d3Network] ===
== gtrendsR - Google Trends ==
* http://christophergandrud.github.io/d3Network/ (old)
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
* https://christophergandrud.github.io/networkD3/ (new)
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
<source lang="rsplus">
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
library(d3Network)
* [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.
Source <- c("A", "A", "A", "A", "B", "B", "C", "C", "D")
* [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.
Target <- c("B", "C", "D", "J", "E", "F", "G", "H", "I")
* [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]
NetworkData <- data.frame(Source, Target)


d3SimpleNetwork(NetworkData, height = 800, width = 1024, file="tmp.html")
== quantmod ==
</source>
[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.


=== [http://cran.r-project.org/web/packages/htmlwidgets/ htmlwidgets for R] ===
# Initial data downloading
Embed widgets in R Markdown documents and Shiny web applications.
# Update existing data
# Create a batch file


* Official website http://www.htmlwidgets.org/.
== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
* [http://deanattali.com/blog/htmlwidgets-tips/ How to write a useful htmlwidgets in R: tips and walk-through a real example]
* 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"]


=== [http://cran.r-project.org/web/packages/networkD3/index.html networkD3] ===
== Tool for connecting Excel with R ==
This is a port of Christopher Gandrud's [http://christophergandrud.github.io/d3Network/ d3Network] package to the htmlwidgets framework.
* 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://cran.r-project.org/web/packages/scatterD3/index.html scatterD3] ===
== write.table ==
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.
=== 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)


=== dygraphs ===
# one liner: row names of a 'matrix' become the names of a vector
* https://github.com/rstudio/dygraphs
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
* [https://www.r-bloggers.com/visualizations-for-algorithmic-trading-in-r/ Visualizations for Algorithmic Trading in R]
all.equal(vec, vec3)
</pre>


=== [https://github.com/bwlewis/rthreejs rthreejs] - Create interactive 3D scatter plots, network plots, and globes ===
=== Avoid leading empty column to header ===
[http://bwlewis.github.io/rthreejs/ Examples]
[https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
 
<pre>
=== d3heatmap ===
write.table(a, 'a.txt', col.names=NA)
See [[Heatmap#d3heatmap|R]]
</pre>
 
=== [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 ===
=== Add blank field AND column names in write.table ===
* [http://moderndata.plot.ly/power-curves-r-plotly-ggplot2/ Power curves] and ggplot2.
* '''write.table'''(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
* [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]
** Suppose x is (n x 2)
* [http://moderndata.plot.ly/filled-chord-diagram-in-r-using-plotly/ Filled chord diagram]
** write.table(x, sep="\t") will generate a file with 2 element on the 1st row
* [https://moderndata.plot.ly/dashboards-in-r-with-shiny-plotly/ DASHBOARDS IN R WITH SHINY & PLOTLY]
** read.table(file) will return an object with a size (n x 2)
* [https://plot.ly/r/shiny-tutorial/ Plotly Graphs in Shiny],
** read.delim(file) and read.delim2(file) will also be correct
** [https://plot.ly/r/shiny-gallery/ Gallery]
* Note that '''write.csv'''() does not have this issue that write.table() has
** [https://plot.ly/r/shinyapp-UN-simple/ Single time series]
** Suppose x is (n x 2)
** [https://plot.ly/r/shinyapp-UN-advanced/ Multiple time series]
** Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
* [https://www.r-exercises.com/2017/09/28/how-to-plot-basic-charts-with-plotly/ How to plot basic charts with plotly]
** If we use read.csv(file), the object is (n x 3). So we need to use '''read.csv(file, row.names = 1)'''
* [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]
* 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]
* [https://blog.methodsconsultants.com/posts/introduction-to-interactive-graphics-in-r-with-plotly/ Introduction to Interactive Graphics in R with plotly]
:<syntaxhighlight lang="rsplus">
 
write.table(a, 'a.txt', col.names=NA)
== 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
* '''readr::write_tsv'''() does not include row names in the output file
<syntaxhighlight lang='rsplus'>
install.packages("devtools")
install.packages("XML")
install.packages("pbapply")
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")
=== read.delim(, row.names=1) and write.table(, row.names=TRUE) ===
# Fetching 30 reviews of 'BOOX Note Ereader,Android 6.0 32 GB 10.3" Dual Touch HD Display'
[https://www.statology.org/read-delim-in-r/ How to Use read.delim Function in R]
#  |++++++++++++++++++++++++++++++++++++++++++++++++++| 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>


== [https://cran.r-project.org/web/packages/gutenbergr/index.html gutenbergr] ==
Case 1: no row.names
[https://blog.jumpingrivers.com/posts/2018/tidytext_edinbr_2018/ Edinbr: Text Mining with R]
<pre>
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>


== Twitter ==
== Read/Write Excel files package ==
[http://www.masalmon.eu/2017/03/19/facesofr/ Faces of #rstats Twitter]
* 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.
== OCR ==
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
* [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?]
** [https://stackoverflow.com/a/17976604 Export both Image and Data from R to an Excel spreadsheet]
* https://cran.r-project.org/web/packages/abbyyR/index.html
* [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.
== Wikipedia ==
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
[https://github.com/ironholds/wikipedir WikipediR]: R's MediaWiki API client library
** [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. 
= Creating local repository for CRAN and Bioconductor =
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
[[R_repository|R repository]]
** [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.
:<syntaxhighlight lang='rsplus'>
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")
</syntaxhighlight>
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R
:<syntaxhighlight lang='rsplus'>
library(writexl)
mylst <- list(sheet1name = df1, sheet2name = df2)
write_xlsx(mylst, "output.xlsx")
</syntaxhighlight>


= Parallel Computing =
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).
See [[R_parallel|R parallel]].
{{Pre}}
> 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
</pre>


= Cloud Computing =
The hidden worksheets become visible (Not sure what are those first rows mean in the output).
 
{{Pre}}
== Install R on Amazon EC2 ==
> excel_sheets("~/Downloads/BRCA.xls")
http://randyzwitch.com/r-amazon-ec2/
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>
 
The Chinese character works too.
{{Pre}}
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</pre>


== Bioconductor on Amazon EC2 ==
To read all worksheets we need a convenient function
http://www.bioconductor.org/help/bioconductor-cloud-ami/
{{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>


= Big Data Analysis =
=== [https://cran.r-project.org/web/packages/readr/ readr] ===
* [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


== bigmemory, biganalytics, bigtabulate ==
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.


== ff, ffbase ==
[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.
* 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 ==
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.
* 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!)


== data.table ==
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.
See [[#data.table_2|data.table]].


= Reproducible Research =
Note that '''data.table::fread()''' can read a selection of the columns.
http://cran.r-project.org/web/views/ReproducibleResearch.html


== Reproducible Environments ==
=== Speed comparison ===
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
[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.


== Some lessons in R coding ==
== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
# 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.
See [[Ggplot2|ggplot2]]
# 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!
 
== Data Manipulation & Tidyverse ==
See [[Tidyverse|Tidyverse]].


= Useful R packages =
== Data Science ==
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
See [[Data_science|Data science]] page
* [https://github.com/qinwf/awesome-R awesome-R]
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]


== Rcpp ==
== microbenchmark & rbenchmark ==
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].
* [https://cran.r-project.org/web/packages/microbenchmark/index.html microbenchmark]
** [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://cran.r-project.org/web/packages/rbenchmark/index.html rbenchmark] (not updated since 2012)


== RInside : embed R in C++ code ==
== 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]]
=== png and resolution ===
It seems people use '''res=300''' as a definition of high resolution.  


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
<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>
</ul>
In the Windows command prompt, run
 
<pre>
=== magick ===
cd C:\R\R-3.0.1\library\RInside\examples\standard
https://cran.r-project.org/web/packages/magick/
make -f Makefile.win
 
</pre>
See an example [[:File:Progpreg.png|here]] I created.
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
 
<pre>
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
rinside_sample0
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
</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
=== geDevices ===
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
* [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.
* http://www.mail-archive.com/rcpp-[email protected]-forge.r-project.org/msg04377.html
* [https://www.jumpingrivers.com/blog/r-knitr-markdown-png-pdf-graphics/ Setting the Graphics Device in a RMarkdown Document]
So the Qt and Wt web tool applications on Windows may or may not be possible.


== GUI ==
=== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ===
=== Qt and R ===
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
* 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 ==
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
On Ubuntu, we need to install tk packages, such as by
<pre>
<pre>
sudo apt-get install tk-dev
sudo apt-get install libgtk2.0-dev libcairo2-dev
</pre>
</pre>


== reticulate - Interface to 'Python' ==
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].
* https://cran.r-project.org/web/packages/reticulate/index.html, [https://github.com/rstudio/reticulate Github]
** Using Python in R markdown
** 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>
=== dpi requirement for publication ===
---
[http://www.cookbook-r.com/Graphs/Output_to_a_file/ For import into PDF-incapable programs (MS Office)]
title: "R Notebook"
output: html_notebook
---


```{r}
=== sketcher: photo to sketch effects ===
library(reticulate)
https://htsuda.net/sketcher/
py_discover_config()
x <- 5
source_python("test.py")
y <- add_three(x)
print(y)
```


Pass R variables to Python. Works
=== httpgd ===
```{python}
* https://nx10.github.io/httpgd/ A graphics device for R that is accessible via network protocols. Display graphics on browsers.
a = 7
* [https://youtu.be/uxyhmhRVOfw Three tricks to make IDEs other than Rstudio better for R development]
print(r.x)
```


Pass python variables to R. Works.
== [http://igraph.org/r/ igraph] ==
```{r}
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]
py$a
py_run_string("y = 10"); py$y
```
</pre>


== Hadoop (eg ~100 terabytes) ==
== Identifying dependencies of R functions and scripts ==
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
 
{{Pre}}
* RHadoop
library(mvbutils)
* Hive
foodweb(where = "package:batr")
* [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)
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
* [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] ===
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
</pre>
* 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.


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


== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
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,
== Colors ==
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
<ul>
<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.
<pre>
<pre>
yum -y install libxml2 libxml2-devel
hcl_palettes(plot = TRUE) # a quick overview
hcl_palettes(palette = "Dark 2", n=5, plot = T)
q4 <- qualitative_hcl(4, palette = "Dark 3")
</pre>
</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]
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.
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.


=== XML ===
[[:File:GgplotPalette.svg]]
* 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()'''.
* 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
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)


# Extract all the paragraphs (HTML tag is p, starting at
=== [http://rpubs.com/gaston/colortools colortools] ===
# the root of the document). Unlist flattens the list to
Tools that allow users generate color schemes and palettes
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


# Replace all by spaces
=== [https://github.com/daattali/colourpicker colourpicker] ===
doc.text = gsub('\n', ' ', doc.text)
A Colour Picker Tool for Shiny and for Selecting Colours in Plots


# Join all the elements of the character vector into a single
=== eyedroppeR ===
# character string, separated by spaces
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]
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.
== [https://github.com/kevinushey/rex rex] ==
<syntaxhighlight lang='rsplus'>
Friendly Regular Expressions
> 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
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
> 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"
</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 ===
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/


== RCurl ==
<pre>
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
library(formatR)
<syntaxhighlight lang='bash'>
tidy_source("Input.R", file = "output.R", width.cutoff=70)
# Test on Ubuntu 14.04
tidy_source("clipboard")  
sudo apt-get install libxml2-dev
# default width is getOption("width") which is 127 in my case.
sudo apt-get install libcurl4-openssl-dev
</pre>
</syntaxhighlight>


=== Scrape google scholar results ===
Some issues
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
 
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] ===
'''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.
<syntaxhighlight lang='bash'>
# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev
 
# Ubuntu 16.04
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev
</syntaxhighlight>
 
[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.
 
=== [https://github.com/hadley/httr httr] ===
httr imports curl, jsonlite, mime, openssl and R6 packages.
 
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:
> tidy_source("clipboard")
* deb: libssl-dev (Debian, Ubuntu, etc)
Error in base::parse(text = code, srcfile = NULL) :
* rpm: openssl-devel (Fedora, CentOS, RHEL)
  3:1: unexpected string constant
* csw: libssl_dev (Solaris)
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
* brew: openssl (Mac OSX)
3: "defg"
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
</pre>
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
* 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.
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
<pre>
--------------------------------------------------------------------
cat("abcd"
ERROR: configuration failed for package ‘openssl’
    ,"defg"  # This is my comment
  ,"ghij")
</pre>
will become
<pre>
cat("abcd", "defg"  # This is my comment
, "ghij")  
</pre>
Still bad!!
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
<pre>
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)
</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.
== styler ==
https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.


=== [http://cran.r-project.org/web/packages/curl/ curl] ===
== Download papers ==
curl is independent of RCurl package.
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
Search and Download Papers from the bioRxiv Preprint Server (biology)


* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
* https://www.opencpu.org/posts/curl-release-0-8/
Interface to the arXiv API
 
=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
* 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 ==
An RStudio addin to run long R commands aside your current session.


<syntaxhighlight lang='rsplus'>
== Teaching ==
library(curl)
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics
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 ===
== Organize R research project ==
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
* [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]


== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
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).
[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]


Example:
=== Naming convention ===
<syntaxhighlight lang='rsplus'>
<ul>
# https://github.com/henrikbengtsson/matrixstats
<li>[https://stackoverflow.com/a/1946879 What is your preferred style for naming variables in R?]
remotes::install_github('HenrikBengtsson/matrixStats@develop')
* 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)
</syntaxhighlight>
* 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


== DirichletMultinomial ==
dataClinicalDesign
On Ubuntu, we do
dataGeneExpression
dataAnnotation
</pre>
<pre>
<pre>
sudo apt-get install libgsl0-dev
# Search all variables ending with .Data
ls()[grep("\\.Data$", ls())]
# Search all variables starting with data_
ls()[grep("^data_", ls())]
</pre>
</pre>
</ul>


== Create GUI ==
=== Efficient Data Management in R ===
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.
 
== Text to speech ==
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
== Speech to text ==
[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://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]


== [http://cran.r-project.org/web/packages/rjson/index.html rjson] ==
== Weather data ==
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/
* [https://github.com/ropensci/prism prism] package
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ==
== logR ==
=== Accessing Bitcoin Data with R ===
https://github.com/jangorecki/logR
http://blog.revolutionanalytics.com/2015/11/accessing-bitcoin-data-with-r.html


=== Plot IP on google map ===
== Progress bar ==
* http://thebiobucket.blogspot.com/2011/12/some-fun-with-googlevis-plotting-blog.html#more  (RCurl, RJONIO, plyr, googleVis)
https://github.com/r-lib/progress#readme
* http://devblog.icans-gmbh.com/using-the-maxmind-geoip-api-with-r/ (RCurl, RJONIO, maps)
* http://cran.r-project.org/web/packages/geoPlot/index.html (geoPlot package (deprecated as 8/12/2013))
* 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.
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'.
<pre>
require(RJSONIO) # fromJSON
require(RCurl)  # getURL


temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
== cron ==
                          ssl.verifypeer = FALSE)
* [https://github.com/bnosac/cronr cronR]
ip <- read.table(textConnection(temp), as.is=TRUE)
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]
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){
== beepr: Play A Short Sound ==
  cat(i, "\n")
https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".
  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
== utils package ==
gmap <- gvisMap(log_gmap, "LatLong",
https://www.rdocumentation.org/packages/utils/versions/3.6.2
                options = list(showTip = TRUE, enableScrollWheel = TRUE,
                              mapType = 'hybrid', useMapTypeControl = TRUE,
                              width = 1024, height = 800))
plot(gmap)
</pre>
[[:File:GoogleVis.png]]


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
== tools package ==
[http://jeffreyhorner.tumblr.com/page/3 Jeffrey Horner's note about deploying Rook App].
* 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>)]


== Map ==
= Different ways of using R =
=== [https://rstudio.github.io/leaflet/ leaflet] ===
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
* 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 ===
== 10 things R can do that might surprise you ==
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
* 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 ===
== R call C/C++ ==
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]
Mainly talks about .C() and .Call().


== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
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.
See an example from [[R#RJSONIO|RJSONIO]] above.


== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
* 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


== gtrendsR - Google Trends ==
=== .Call ===
* [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/base/versions/3.6.2/topics/CallExternal ?.Call]
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
* [http://mazamascience.com/WorkingWithData/?p=1099 Using R — .Call(“hello”)]
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
* http://adv-r.had.co.nz/C-interface.html
* [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://working-with-data.mazamascience.com/2021/07/16/using-r-callhello/ Using R – .Call(“hello”)]
* [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 ==
Be sure to add the ''PACKAGE'' parameter to avoid an error like
[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.
<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>


# Initial data downloading
=== NAMESPACE file & useDynLib ===
# Update existing data
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
# Create a batch file
* 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]


== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
(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
* 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
library.dynam("libname", package, lib.loc)
* https://github.com/cran/caret source code mirrored on github
</pre>
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/


== Tool for connecting Excel with R ==
=== library.dynam.unload() ===
* https://bert-toolkit.com/
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
* 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.
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]


== Read/Write Excel files package ==
=== gcc ===
* http://www.milanor.net/blog/?p=779
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
* [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/xlsx/index.html xlsx]: depends on Java
=== Primitive functions ===
* [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.
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]
** 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'>
== SEXP ==
library(readxl)
Some examples from packages
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).
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
<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.
== R call Fortran ==
<syntaxhighlight lang='rsplus'>
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
> read_excel("~/Downloads/testChinese.xlsx", 1)
* [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)
  中文 B C
1     a b c
2    1 2 3
</syntaxhighlight>


To read all worksheets we need a convenient function
== Embedding R ==
<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) ==
* 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


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.
=== 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>.


[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released.
This example can be run by
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</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.
Note:
# '''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.
# '''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''.


Note that '''fread()''' can read-n a selection of the columns.
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.


== Colors ==
=== An example from Bioconductor workshop ===
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* [http://colorspace.r-forge.r-project.org/articles/colorspace.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes.
* 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://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* 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://stackoverflow.com/questions/7457635/calling-r-function-from-c
* [http://www.ucl.ac.uk/~zctpep9/Archived%20webpages/Cookbook%20for%20R%20%C2%BB%20Colors%20(ggplot2).htm 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
** RColorBrewer::display.brewer.all()
** For example, [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/
* [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.  
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>
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


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.
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
</pre>


[[:File:GgplotPalette.svg]]
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].
<pre>
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.
</pre>


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


== Data Manipulation & Tidyverse ==
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
See [[Tidyverse|Tidyverse]].


== Data Science ==
=== Create a Simple Socket Server in R ===
See [[Data_science|Data science]] page
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].


== microbenchmark & rbenchmark ==
Create an R function
* [https://cran.r-project.org/web/packages/microbenchmark/index.html microbenchmark]
** [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://cran.r-project.org/web/packages/rbenchmark/index.html rbenchmark] (not updated since 2012)
 
== [http://cran.r-project.org/web/packages/jpeg/index.html 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 [http://moderndata.plot.ly/fun-with-heatmaps-and-plotly/ Fun with Heatmaps and Plotly].
 
== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ==
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
 
== [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
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 '^]'.


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].
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                 


== [http://igraph.org/r/ igraph] ==
R> quit
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]
Connection closed by foreign host.
</pre>


== Identifying dependencies of R functions and scripts ==
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
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]].
<syntaxhighlight lang='rsplus'>
library(mvbutils)
foodweb(where = "package:batr")


foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
See my [[Rserve]] page.
 
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
</syntaxhighlight>


== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
=== outsider ===
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
* [https://joss.theoj.org/papers/10.21105/joss.02038 outsider]: Install and run programs, outside of R, inside of R
* [https://github.com/stephenturner/om..bcftools Run bcftools with outsider in R]


Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===


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


=== [http://rpubs.com/gaston/colortools colortools] ===
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
Tools that allow users generate color schemes and palettes
* [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://github.com/daattali/colourpicker colourpicker] ===
Terminal
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
{{Pre}}
# 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
</pre>
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'''


=== [https://cran.r-project.org/web/packages/inlmisc/index.html inlmisc] ===
Now go back to R
[https://owi.usgs.gov/blog/tolcolors/ GetTolColors()]. Lots of examples.
{{Pre}}
install.packages("rJava")
</pre>
Done!


== [https://github.com/kevinushey/rex rex] ==
If above does not work, a simple way is by (under Ubuntu) running
Friendly Regular Expressions
<pre>
sudo apt-get install r-cran-rjava
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
=== RCaller ===
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''


See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
=== RApache ===
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf


=== Rscript, arguments and commandArgs() ===
[https://www.r-bloggers.com/passing-arguments-to-an-r-script-from-command-lines/ Passing arguments to an R script from command lines]
Syntax:
<pre>
<pre>
library(formatR)
$ Rscript --help
tidy_source("Input.R", file = "output.R", width.cutoff=70)
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
tidy_source("clipboard")
# default width is getOption("width") which is 127 in my case.
</pre>
</pre>


Some issues
Example:
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
<pre>
<pre>
cat("abcd",
args = commandArgs(trailingOnly=TRUE)
    # This is my comment
# test if there is at least one argument: if not, return an error
    "defg")
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")
</pre>
</pre>
will result in
<pre>
<pre>
> tidy_source("clipboard")
Rscript --vanilla sillyScript.R iris.txt out.txt
Error in base::parse(text = code, srcfile = NULL) :
# args[1] =  iris.txt
  3:1: unexpected string constant
# args[2] =  out.txt
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
</pre>
3: "defg"
 
  ^
=== Rscript, #! Shebang and optparse package ===
</pre>
<ul>
* 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.  
<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>
<pre>
cat("abcd"
#!/usr/bin/env Rscript
    ,"defg"  # This is my comment
print ("shebang works")
  ,"ghij")
</pre>
</pre>
will become
Then in the command line
<pre>
<pre>
cat("abcd", "defg"  # This is my comment
chmod u+x shebang.R
, "ghij")
./shebang.R
</pre>
</pre>
Still bad!!
<li>[http://www.cureffi.org/2014/01/15/running-r-batch-mode-linux/ Running R in batch mode on Linux]
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
<li>[https://cran.r-project.org/web/packages/optparse/index.html optparse] package. Check out its vignette.
<pre>
<li>[https://cran.r-project.org/web/packages/getopt/index.html getopt]: C-Like 'getopt' Behavior.
cat("</p>",
</ul>
"<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
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),
Provides hash-bang (#!) capability for R
file=ExternalFileName, sep="\n", append=T)
 
</pre>
FAQs:
will result in
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
<pre>
* [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]
> tidy_source("clipboard", width.cutoff=70)
* [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?]
Error in base::parse(text = code, srcfile = NULL) :
{{Pre}}
  3:129: unexpected SPECIAL
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
# R 3.5.2 docker container
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
</pre>
                                              # Example: r --verbose -e "date()"
* ''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>
-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",  
                                              # Example: R -q -e "date()"
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,
 
    sep = "\n", append = T)
-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
</pre>
                                              # It won't show the startup message when it is used in the command line.
* It merges lines though I don't always want to do that. For example
                                              # Example: Rscript -e "date()"
<pre>
cat("abcd"
    ,"defg" 
  ,"ghij")
</pre>
will become
<pre>
cat("abcd", "defg", "ghij")
</pre>
</pre>


== Download papers ==
We can install littler using two ways.
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
* 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.
Search and Download Papers from the bioRxiv Preprint Server
* 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.
 
'''r''' was not meant to run interactively like '''R'''. See ''man r''.


=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
=== RInside: Embed R in C++ ===
Interface to the arXiv API
See [[R#RInside|RInside]]


=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
(''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.
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
* 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 ==
The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
An RStudio addin to run long R commands aside your current session.


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


== Organize R research project ==
To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>
* [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())'''.
** 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.''
** No update for the ''here'' package. 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]


=== How to save (and load) datasets in R (.Rdata vs .Rds file) ===
The real build process looks like (check <Makefile> for completeness)
[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]
<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>


== Text to speech ==
Hello World example of embedding R in C++.
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]
<pre>
#include <RInside.h>                    // for the embedded R via RInside


== Weather data ==
int main(int argc, char *argv[]) {
* [https://github.com/ropensci/prism prism] package
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== logR ==
    RInside R(argc, argv);              // create an embedded R instance
https://github.com/jangorecki/logR


== Progress bar ==
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
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'.
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns


== cron ==
    exit(0);
[https://github.com/bnosac/cronr cronR]
}
</pre>


= Different ways of using R =
The above can be compared to the Hello world example in Qt.
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
<pre>
#include <QApplication.h>
#include <QPushButton.h>


== 10 things R can do that might surprise you ==
int main( int argc, char **argv )
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
{
    QApplication app( argc, argv );


== R call C/C++ ==
    QPushButton hello( "Hello world!", 0 );
Mainly talks about .C() and .Call().
    hello.resize( 100, 30 );


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.
    app.setMainWidget( &hello );
    hello.show();


* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
    return app.exec();
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
}
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
</pre>
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
* http://www.stat.harvard.edu/ccr2005/
* http://mazamascience.com/WorkingWithData/?p=1099


=== NAMESPACE file & useDynLib ===
=== [http://www.rfortran.org/ RFortran] ===
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
RFortran is an open source project with the following aim:
* 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]


(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
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
<syntaxhighlight lang='rsplus'>
library.dynam("libname", package, lib.loc)
</syntaxhighlight>


=== library.dynam.unload() ===
It works only on Windows platform with Microsoft Visual Studio installed:(
* 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]


=== gcc ===
== Call R from other languages ==
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
=== C ===
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


== SEXP ==
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]
Some examples from packages


* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace


== R call Fortran ==
=== JRI ===
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
http://www.rforge.net/JRI/
* [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)


== Embedding R ==
=== ryp2 ===
http://rpy.sourceforge.net/rpy2.html


* 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.
== Create a standalone Rmath library ==
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
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].
* [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


=== An very simple example (do not return from shell) from Writing R Extensions manual ===
Here is my experience based on R 3.0.2 on Windows OS.
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
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</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.
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


Note:
=== Use Rmath library in our code ===
# '''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''.
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.


More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
# 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.


=== An example from Bioconductor workshop ===
# Force to link against the static library <libRmath.a>
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
* 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].
# OR
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c


Example:
# Force to link against dynamic library <Rmath.dll>
Create <embed.c> file
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
</pre>
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!
<pre>
<pre>
#include <Rembedded.h>
c:\R>RmathEx1
#include <Rdefines.h>
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>


static void doSplinesExample();
Below is the cpp program <RmathEx1.cpp>.
int
<pre>
main(int argc, char *argv[])
//RmathEx1.cpp
{
#define MATHLIB_STANDALONE
    Rf_initEmbeddedR(argc, argv);
#include <iostream>
    doSplinesExample();
#include "Rmath.h"
    Rf_endEmbeddedR(0);
    return 0;
}
static void
doSplinesExample()
{
    SEXP e, result;
    int errorOccurred;


    // create and evaluate 'library(splines)'
using std::cout; using std::cin; using std::endl;
    PROTECT(e = lang2(install("library"), mkString("splines")));
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
    if (errorOccurred) {
        // handle error
    }
    UNPROTECT(1);


    // 'options(FALSE)' ...
int main()
    PROTECT(e = lang2(install("options"), ScalarLogical(0)));
{
    // ... modified to 'options(example.ask=FALSE)' (this is obscure)
  double x1, x2;
    SET_TAG(CDR(e), install("example.ask"));
  cout << "Enter a argument for the normal cdf:" << endl;
    R_tryEval(e, R_GlobalEnv, NULL);
  cin >> x1;
    UNPROTECT(1);
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


    // 'example("ns")'
  cout << "Prob(Z <= " << x1 << ") = " <<
     PROTECT(e = lang2(install("example"), mkString("ns")));
     pnorm(x1, 0, 1, 1, 0)  << endl;
     R_tryEval(e, R_GlobalEnv, &errorOccurred);
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    UNPROTECT(1);
     pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
}
</pre>
</pre>
Then build the executable. Note that I don't need to create R_HOME variable.
<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
== Calling R.dll directly ==
# Using a single line will give an error and cannot not show the real problem.
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.
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
 
# A better way is to run compile and link separately
== Create HTML report ==
gcc -I../../include -c embed.c
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].
gcc -o embed embed.o -L../../lib -lR -lRblas
 
../../bin/R CMD ./embed
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
</pre>
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.
 
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
* [http://gforge.se/2020/07/news-in-htmltable-2-0/ News in htmlTable 2.0]


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].
=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
<pre>
* https://github.com/renkun-ken/formattable
export R_HOME=/home/brb/Downloads/R-3.0.2
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]
./embed # No need to include R CMD in front.
</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].
=== [https://github.com/crubba/htmltab htmltab] package ===
This package is NOT used to CREATE html report but EXTRACT html table.


Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
=== [http://cran.r-project.org/web/packages/ztable/index.html 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 a Simple Socket Server in R ===
== Create academic report ==
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].  
[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.


Create an R function
== Create pdf and epub files ==
<pre>
{{Pre}}
simpleServer <- function(port=6543)
# Idea:
{
#        knitr        pdflatex
  sock <- socketConnection ( port=port , server=TRUE)
#   rnw -------> tex ----------> pdf
  on.exit(close( sock ))
library(knitr)
   cat("\nWelcome to R!\nR>" ,file=sock )
knit("example.rnw") # create example.tex file
  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
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
<pre>
{{Pre}}
$ telnet localhost 6543
git clone https://github.com/yihui/knitr-examples.git
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.
</pre>
</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!


=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
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.
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]].


See my [[Rserve]] page.
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


=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
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 !!
</pre>


=== [http://rdotnet.codeplex.com/ R.NET] ===
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
{{Pre}}
# Idea:
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
library(knitr)
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
knit("DESeq2.Rnw") # create DESeq2.tex
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
 
Terminal
<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>
</pre>
* And then run '''sudo ldconfig'''


Now go back to R
Convert tex to epub
<syntaxhighlight lang='rsplus'>
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
install.packages("rJava")
</syntaxhighlight>
Done!


If above does not work, a simple way is by (under Ubuntu) running
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
<pre>
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
sudo apt-get install r-cran-rjava
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


=== RCaller ===
* 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


=== RApache ===
== Create Word report ==
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf


=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
=== Using the power of Word ===
Provides hash-bang (#!) capability for R
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]


FAQs:
=== knitr + pandoc ===
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* [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]
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* [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?]
* http://rmarkdown.rstudio.com/articles_docx.html
<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
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.
                                              # Example: R -q -e "date()"
<pre>
# Idea:
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</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.


-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
Another way is
                                              # It won't show the startup message when it is used in the command line.
<pre>
                                              # Example: Rscript -e "date()"
library(pander)
</syntaxhighlight>
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
</pre>


We can install littler using two ways.
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:
* 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.
* A pdf file: pandoc -s report.md -t latex -o report.pdf
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.
* 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


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.
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>
knit("example.Rmd")
pandoc("example.md", format="epub")
</pre>


'''r''' was not meant to run interactively like '''R'''. See ''man r''.
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<pre>
> 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
</pre>


=== RInside: Embed R in C++ ===
=== pander ===
See [[R#RInside|RInside]]
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:


(''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.
<pre>
library(pander)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
            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.


The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
# http://johnmacfarlane.net/pandoc/
# http://rapporter.github.com/pander/
# http://rapporter.github.com/pander/#examples


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'.
=== 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>
> library(R2wd)
> wdGet()
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
    installstatconnDCOM()


To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
This will download and install the current version of statconnDCOM
<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)
You will need a working Internet connection
<pre>
because installation needs to download a file.
g++ -I/home/brb/Downloads/R-3.0.2/include \
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
   argument is of length zero
    -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>
</pre>


Hello World example of embedding R in C++.
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
<pre>
#include <RInside.h>                    // for the embedded R via RInside


int main(int argc, char *argv[]) {
=== 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


    RInside R(argc, argv);              // create an embedded R instance
=== rtf ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
Package xtable will produce html output.
{{Pre}}
print(xtable(X), type="html")
</pre>
 
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.


    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
=== officer ===
<ul>
<li>[https://cran.r-project.org/web/packages/officer/index.html CRAN]. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.  
<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.
<li>Add a line space
<pre>
doc <- body_add_par(doc, "")


     exit(0);
# 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)
</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>
</pre>
</ul>


The above can be compared to the Hello world example in Qt.
== Powerpoint ==
<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>
<pre>
#include <QApplication.h>
library(gridExtra)
#include <QPushButton.h>
grid.newpage()
grid.table(mydf)
</pre>
</li>
<li>[https://bookdown.org/yihui/rmarkdown/powerpoint-presentation.html Rmarkdown]
</li>
</ul>


int main( int argc, char **argv )
== PDF manipulation ==
{
[https://github.com/pridiltal/staplr staplr]
    QApplication app( argc, argv );


    QPushButton hello( "Hello world!", 0 );
== R Graphs Gallery ==
    hello.resize( 100, 30 );
* [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]


    app.setMainWidget( &hello );
== COM client or server ==
    hello.show();


    return app.exec();
=== Client ===
}
</pre>


=== [http://www.rfortran.org/ RFortran] ===
[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
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.''
=== Server ===
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


It works only on Windows platform with Microsoft Visual Studio installed:(
== Use R under proxy ==
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


== Call R from other languages ==
== RStudio ==
=== C ===
* [https://github.com/rstudio/rstudio Github]
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]
* 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]
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]
 
=== rstudio.cloud ===
https://rstudio.cloud/
 
=== Launch RStudio ===
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]


Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
=== 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.


=== JRI ===
With an RStudio project file, you can
http://www.rforge.net/JRI/
* 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


=== ryp2 ===
=== package search ===
http://rpy.sourceforge.net/rpy2.html
https://github.com/RhoInc/CRANsearcher


== Create a standalone Rmath library ==
=== Git ===
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].
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]


Here is my experience based on R 3.0.2 on Windows OS.
== 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]


=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
== List files using regular expression ==
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.
* Extension
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
list.files(pattern = "^Something")
make -f Makefile.win
</pre>
</pre>


=== Use Rmath library in our code ===
Using '''Sys.glob()"' as
<pre>
<pre>
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
> Sys.glob("~/Downloads/*.txt")
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
# It is not LD_LIBRARY_PATH in above.
</pre>


# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
== Hidden tool: rsync in Rtools ==
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
<pre>
# It is OK to save the cpp file under any directory.
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe


# Force to link against the static library <libRmath.a>
sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
total size is 1198416  speedup is 3.71
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


# Force to link against dynamic library <Rmath.dll>
c:\Rtools\bin>
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
</pre>
</pre>
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!
 
<pre>
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].
c:\R>RmathEx1
 
Enter a argument for the normal cdf:
== Install rgdal package (geospatial Data) on ubuntu ==
1
Terminal
Enter a argument for the chi-squared cdf:
{{Pre}}
1
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
Prob(Z <= 1) = 0.841345
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
Prob(Chi^2 <= 1)= 0.682689
</pre>
</pre>


Below is the cpp program <RmathEx1.cpp>.
R
<pre>
{{Pre}}
//RmathEx1.cpp
install.packages("rgdal")
#define MATHLIB_STANDALONE
</pre>
#include <iostream>
#include "Rmath.h"


using std::cout; using std::cin; using std::endl;
== Install sf package ==
I got the following error even I have installed some libraries.
<pre>
checking GDAL version >= 2.0.1... no
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


int main()
sudo apt update
{
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
  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 << ") = " <<
sudo apt install libgdal-dev # works on ubuntu 20.04 too
    pnorm(x1, 0, 1, 1, 0)  << endl;
                            # no need the previous lines
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
</pre>
</pre>


== Calling R.dll directly ==
== Database ==
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://cran.r-project.org/web/views/Databases.html
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]


== Create HTML report ==
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor.
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
'''Creating a new database''':
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}}
library(DBI)


* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
dbDisconnect(mydb)
unlink("my-db.sqlite")


=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
# temporary database
* https://github.com/renkun-ken/formattable
mydb <- dbConnect(RSQLite::SQLite(), "")
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
dbDisconnect(mydb)
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]
</pre>


=== [https://github.com/crubba/htmltab htmltab] package ===
'''Loading data''':
This package is NOT used to CREATE html report but EXTRACT html table.
{{Pre}}
mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)


=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
dbListTables(mydb)
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 ==
dbListFields(con, "mtcars")
[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.


== Create pdf and epub files ==
dbReadTable(con, "mtcars")
<syntaxhighlight lang='rsplus'>
</pre>
# 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.
'''Queries''':
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
<syntaxhighlight lang='bash'>
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf


git clone https://github.com/yihui/knitr-examples.git
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</syntaxhighlight>


To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
<syntaxhighlight lang='rsplus'>
dbFetch(res)
# Idea:
</pre>
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


library(knitr)
'''Batched queries''':
knit("DESeq2.Rnw") # create DESeq2.tex
{{Pre}}
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
dbClearResult(rs)
</syntaxhighlight>
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
<pre>
while (!dbHasCompleted(rs)) {
## Windows OS, epub cannot be built
  df <- dbFetch(rs, n = 10)
pandoc:
  print(nrow(df))
Error:
}
"source" (line 41, column 7):
unexpected "k"
expecting "{document}"


## Linux OS, epub missing figures and R codes.
dbClearResult(rs)
## 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
'''Multiple parameterised queries''':
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
{{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>


=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
'''Statements''':
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
{{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://rmarkdown.rstudio.com/lesson-7.html
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
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]
* 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 ==
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===


=== knitr + pandoc ===
=== [[MySQL#Use_through_R|RMySQL]] ===
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.
* 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.
=== MongoDB ===
<pre>
* http://www.r-bloggers.com/r-and-mongodb/
# Idea:
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</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
=== odbc ===
<pre>
library(pander)
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
</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:
=== RODBC ===
* 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!
=== DBI ===
<pre>
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)
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
<pre>
* To use databases with dplyr, you need to first install dbplyr
> pandoc("Rmd_to_Epub.md", format="epub")
* https://db.rstudio.com/dplyr/
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html
Error in (function (input, format, ext, cfg) : conversion failed
 
In addition: Warning message:
'''Create a new SQLite database''':
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
{{Pre}}
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
</pre>
</pre>


=== pander ===
'''Connect to a database''':
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}}
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


<pre>
library(dbplyr)
library(pander)
library(dplyr)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
mammals <- src_sqlite("portal_mammals.sqlite")
            output = tempfile(), convert = 'docx')
</pre>
</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/
'''Querying the database with the SQL syntax''':
# http://rapporter.github.com/pander/
{{Pre}}
# http://rapporter.github.com/pander/#examples
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
</pre>


=== R2wd ===
'''Querying the database with the dplyr syntax''':
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>
surveys <- tbl(mammals, "surveys")
> library(R2wd)
surveys %>%
> wdGet()
    select(year, species_id, plot_id)
Loading required package: rcom
head(surveys, n = 10)
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
show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
</pre>


You will need a working Internet connection
'''Simple database queries''':
because installation needs to download a file.
{{Pre}}
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
surveys %>%
  argument is of length zero
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</pre>
</pre>


The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
'''Laziness''' (instruct R to stop being lazy):
{{Pre}}
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</pre>


=== Convert from pdf to word ===
'''Complex database queries''':
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert
{{Pre}}
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table


=== rtf ===
surveys # The plot_id column also features in the surveys table
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
# Join databases method 1
Package xtable will produce html output. <syntaxhighlight lang='rsplus'>print(xtable(X), type="html")</syntaxhighlight>  
plots %>%
 
  filter(plot_id == 1) %>%
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.
  inner_join(surveys) %>%
  collect()
</pre>


=== [http://cran.r-project.org/web/packages/ReporteRs/index.html ReporteRs] ===
=== NoSQL ===
Microsoft Word, Microsoft Powerpoint and HTML documents generation from R. The source code is hosted on https://github.com/davidgohel/ReporteRs
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]


[https://statbandit.wordpress.com/2016/10/28/a-quick-exploration-of-reporters/ A quick exploration]
== Github ==


== PDF manipulation ==
=== R source  ===
[https://github.com/pridiltal/staplr staplr]
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.


== R Graphs Gallery ==
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
* [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.


== COM client or server ==
=== R packages (only) source (metacran) ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.


=== Client ===
=== Bioconductor packages source ===
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>


[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
=== Send local repository to Github in R by using reports package ===
http://www.youtube.com/watch?v=WdOI_-aZV0Y


=== Server ===
=== My collection ===
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]
* 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


== Use R under proxy ==
=== How to download ===
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


== RStudio ==
Clone ~ Download.
* [https://github.com/rstudio/rstudio Github]
* Command line
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
<pre>
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]
git clone https://gist.github.com/4484270.git
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


=== rstudio.cloud ===
* Within R
https://rstudio.cloud/
<pre>
library(devtools)
source_gist("4484270")
</pre>
or
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
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)
})
</pre>


=== Launch RStudio ===
=== Jekyll ===
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]


=== Create .Rproj file ===
== Connect R with Arduino ==
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://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


With an RStudio project file, you can
== Android App ==
* Restore .RData into workspace at startup
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
* Save workspace to .RData on exit
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
* Always save history (even if no saving .RData)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]
* etc


=== package search ===
== Common plots tips ==
https://github.com/RhoInc/CRANsearcher
=== Create an empty plot ===
'''plot.new()'''   


=== Git ===
=== Overlay plots ===
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
[https://finnstats.com/index.php/2021/08/15/how-to-overlay-plots-in-r/ How to Overlay Plots in R-Quick Guide with Example].
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]
<pre>
#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)
</pre>


== Visual Studio ==
=== Save the par() and restore it ===
[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]
'''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.
== List files using regular expression ==
* 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'.
* Extension
<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>
<pre>
list.files(pattern = "\\.txt$")
ex <- function() {
  old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
  on.exit(par(old.par))
  ## ... do lots of par() settings and plots
  ## ...
  invisible() #-- now,  par(old.par) will be executed
}
</pre>
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
'''Example 3''': It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.
* Start with
<pre>
<pre>
list.files(pattern = "^Something")
ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
</pre>
</pre>
Using '''Sys.glob()"' as
<pre>
<pre>
> Sys.glob("~/Downloads/*.txt")
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
ex()
par()$mar
</pre>
</pre>


== Hidden tool: rsync in Rtools ==
=== Grouped boxplots ===
<pre>
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
** 'at' parameter in boxplot() to change the equal spaced boxplots
sending incremental file list
** embed par(mar=) in boxplot()
a.exe
** 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)


sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
=== [https://www.samruston.co.uk/ Weather Time Line] ===
total size is 1198416  speedup is 3.71
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].


c:\Rtools\bin>
=== Horizontal bar plot ===
{{Pre}}
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") 
</pre>
</pre>
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
[[:File:Ggplot2bar.svg]]
total size is 8036311  speedup is 1.95


c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
=== Include bar values in a barplot ===
cygwin warning:
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
  MS-DOS style path detected: c:\users\limingc\Documents\binary\procexp.exe
* [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.
  Preferred POSIX equivalent is: /cygdrive/c/users/limingc/Documents/binary/procexp.exe
* [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]
  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/"
Use text().
sending incremental file list
 
binary/
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].
binary/procexp.exe


sent 1767277 bytes  received 35 bytes  3534624.00 bytes/sec
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.
total size is 8036311  speedup is 4.55


c:\Rtools\bin>
=== Grouped barplots ===
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
{{Pre}}
library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
      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)


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
=== Math expression ===
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* Some cases
** 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")


== Install rgdal package (geospatial Data) on ubuntu ==
# Superscript
Terminal
plot(1:10, main = expression("My Title"^2))  
<syntaxhighlight lang='bash'>
# Subscript
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
plot(1:10, main = expression("My Title"[2])) 
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
</syntaxhighlight>


R
# Expressions with Spacing
<syntaxhighlight lang='rsplus'>
# '~' is to add space and '*' is to squish characters together
install.packages("rgdal")
plot(1:10, xlab= expression(Delta * 'C'))
</syntaxhighlight>
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")


== Install sf package ==
# Substituting Expressions
I got the following error even I have installed some libraries.
plot(x,y,
<pre>
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
checking GDAL version >= 2.0.1... no
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
configure: error: sf is not compatible with GDAL versions below 2.0.1
    main = "Substituted Expressions")
</pre>
</pre>
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
=== Impose a line to a scatter plot ===
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
* abline + lsfit # least squares
sudo apt install libgdal-dev
{{Pre}}
</syntaxhighlight>
plot(cars)
 
abline(lsfit(cars[, 1], cars[, 2]))
== Set up Emacs on Windows ==
# OR
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
abline(lm(cars[,2] ~ cars[,1]))
<pre>
</pre>
(setq-default inferior-R-program-name
* abline + line # robust line fitting
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
{{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>
</pre>


== Database ==
=== How to actually make a quality scatterplot in R: axis(), mtext() ===
* https://cran.r-project.org/web/views/Databases.html
[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]
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]
 
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite
 
'''Creating a new database''':
<syntaxhighlight lang='rsplus'>
library(DBI)


mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
=== 3D scatterplot ===
dbDisconnect(mydb)
* [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'''.
unlink("my-db.sqlite")
* [[R_web#plotly|R web > plotly]]


# temporary database
=== Rotating x axis labels for barplot ===
mydb <- dbConnect(RSQLite::SQLite(), "")
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
dbDisconnect(mydb)
{{Pre}}
</syntaxhighlight>
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</pre>


'''Loading data''':
=== Set R plots x axis to show at y=0 ===
<syntaxhighlight lang='rsplus'>
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
mydb <- dbConnect(RSQLite::SQLite(), "")
{{Pre}}
dbWriteTable(mydb, "mtcars", mtcars)
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
dbWriteTable(mydb, "iris", iris)
</pre>


dbListTables(mydb)
=== Different colors of axis labels in barplot ===
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]


dbListFields(con, "mtcars")
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
{{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>


dbReadTable(con, "mtcars")
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
</syntaxhighlight>
{{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>
 
=== Use text() to draw labels on X/Y-axis including rotation ===
* 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]
* [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/
{{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/


'''Queries''':
=== Vertically stacked plots with the same x axis ===
<syntaxhighlight lang='rsplus'>
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
=== Include labels on the top axis/margin: axis() and mtext() ===
<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]]


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
This can be used to annotate each plot with the script name, date, ...
<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>


res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
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].
dbFetch(res)
</syntaxhighlight>


'''Batched queries''':
=== Legend tips ===
<syntaxhighlight lang='rsplus'>
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


dbClearResult(rs)
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
</syntaxhighlight>
{{Pre}}
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)
</pre>


'''Multiple parameterised queries''':
'''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].
<syntaxhighlight lang='rsplus'>
<pre>
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
legend("bottomright", inset=.05, )
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
</pre>
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)
</syntaxhighlight>


'''Statements''':
'''legend without a box'''
<syntaxhighlight lang='rsplus'>
<pre>
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
legend(, bty = "n")
#> [1] 0
</pre>
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</syntaxhighlight>


=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
'''Add a legend title'''
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]
<pre>
legend(, title = "")
</pre>


=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.


=== [[MySQL#Use_through_R|RMySQL]] ===
=== Superimpose a density plot or any curves ===
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
Use '''lines()'''.  
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.


=== MongoDB ===
Example 1
* http://www.r-bloggers.com/r-and-mongodb/
{{Pre}}
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))


=== odbc ===
plot(density(x), col = "#6F69AC", lwd = 3)
lines(density(y), col = "#95DAC1", lwd = 3)
lines(density(z), col = "#FFEBA1", lwd = 3)
</pre>


=== RODBC ===
Example 2
 
{{Pre}}
=== DBI ===
require(survival)
 
n = 10000
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
beta1 = 2; beta2 = -1
* To use databases with dplyr, you need to first install dbplyr
lambdaT = 1 # baseline hazard
* https://db.rstudio.com/dplyr/
lambdaC = 2  # hazard of censoring
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
set.seed(1234)
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html
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>


'''Create a new SQLite database''':
Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".
<syntaxhighlight lang='rsplus'>
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")


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


copy_to(my_db, surveys)
[[:File:Logscale.png]]
copy_to(my_db, plots)
my_db
</syntaxhighlight>


'''Connect to a database''':
=== Custom scales ===
<syntaxhighlight lang='rsplus'>
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


library(dbplyr)
== Time series ==
library(dplyr)
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
mammals <- src_sqlite("portal_mammals.sqlite")
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
</syntaxhighlight>


'''Querying the database with the SQL syntax''':
=== Time series stock price plot ===
<syntaxhighlight lang='rsplus'>
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
* [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]
</syntaxhighlight>
* https://timelyportfolio.github.io/rCharts_time_series/history.html


'''Querying the database with the dplyr syntax''':
{{Pre}}
<syntaxhighlight lang='rsplus'>
library(quantmod)
surveys <- tbl(mammals, "surveys")
getSymbols("AAPL")
surveys %>%
getSymbols("IBM") # similar to AAPL
    select(year, species_id, plot_id)
getSymbols("CSCO") # much smaller than AAPL, IBM
head(surveys, n = 10)
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')


show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
tail(Cl(DJI))
</syntaxhighlight>
</pre>


'''Simple database queries''':
=== tidyquant: Getting stock data ===
<syntaxhighlight lang='rsplus'>
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</syntaxhighlight>


'''Laziness''' (instruct R to stop being lazy):
=== Timeline plot ===
<syntaxhighlight lang='rsplus'>
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
data_subset <- surveys %>%
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</syntaxhighlight>


'''Complex database queries''':
=== Clockify ===
<syntaxhighlight lang='rsplus'>
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]
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
== Circular plot ==
* http://freakonometrics.hypotheses.org/20667 which uses [https://cran.r-project.org/web/packages/circlize/ circlize] package; see also the '''ComplexHeatmap''' package.
* https://www.biostars.org/p/17728/
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.


# Join databases method 1
== Word cloud ==
plots %>%
* [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]
  filter(plot_id == 1) %>%
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
  inner_join(surveys) %>%
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
  collect()
* [https://github.com/lepennec/ggwordcloud?s=09 ggwordcloud]
</syntaxhighlight>


=== NoSQL ===
== Text mining ==
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
* [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].


== Github ==
== 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)


=== R source  ===
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]


If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
* [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]


=== R packages (only) source (metacran) ===
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams


=== Bioconductor packages source ===
=== DAGitty (browser-based and R package) ===
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
* http://dagitty.net/
* https://cran.r-project.org/web/packages/dagitty/index.html


=== Send local repository to Github in R by using reports package ===
=== dagR ===
http://www.youtube.com/watch?v=WdOI_-aZV0Y
* https://cran.r-project.org/web/packages/dagR


=== My collection ===
=== Gmisc ===
* https://github.com/arraytools
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]
* 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 ===
=== Concept Maps ===
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.


Clone ~ Download.
=== flow ===
* Command line
[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]
<pre>
git clone https://gist.github.com/4484270.git
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


* Within R
== Venn Diagram ==
<pre>
[[Venn_diagram|Venn diagram]]
library(devtools)
source_gist("4484270")
</pre>
or
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
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)
})
</pre>


=== Jekyll ===
== hexbin plot ==
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
* [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.


== Connect R with Arduino ==
== Bump chart/Metro map ==
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
https://dominikkoch.github.io/Bump-Chart/
* 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 ==
== Amazing/special plots ==
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
See [[Amazing_plot|Amazing plot]].
* [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 ==
== Google Analytics ==
=== Grouped boxplots ===
=== GAR package ===
* [http://sphaerula.com/legacy/R/boxplotTwoWay.html Box Plots of Two-Way Layout]
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html
* [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] ===
== Linear Programming ==
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].
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/
 
== Linear Algebra ==
* [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.
 
== Amazon Alexa ==
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
 
== 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


=== Horizontal bar plot ===
== Secure API keys ==
<syntaxhighlight lang='rsplus'>
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
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") 
</syntaxhighlight>


[[:File:Ggplot2bar.svg]]
== Credentials and secrets ==
[https://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]


=== Include bar values in a barplot ===
== Hide a password ==
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
=== keyring package ===
* [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://cran.r-project.org/web/packages/keyring/index.html
* [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]
* [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]


Use text().  
=== getPass ===
[https://cran.r-project.org/web/packages/getPass/README.html getPass]


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].
== Vision and image recognition ==
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
* [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


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.
== Creating a Dataset from an Image ==
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]


=== Grouped barplots ===
== Turn pictures into coloring pages ==
* 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://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
library(ggplot2)
 
# mydata <- data.frame(OUTGRP, INGRP, value)
== Numerical optimization ==
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
[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]
      geom_bar(position="dodge", stat="identity")
 
</syntaxhighlight>
* [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]
* 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'>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
> 1 - 2*(1-pnorm(1))
* [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.  
[1] 0.6826895
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
> 1 - 2*(1-pnorm(1.96))
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
[1] 0.9500042
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
</syntaxhighlight>
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.
* [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 ===
== Ryacas: R Interface to the 'Yacas' Computer Algebra System ==
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
[https://blog.ephorie.de/doing-maths-symbolically-r-as-a-computer-algebra-system-cas Doing Maths Symbolically: R as a Computer Algebra System (CAS)]
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* https://andyphilips.github.io/blog/2017/08/16/mathematical-symbols-in-r-plots.html
** 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
** 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)
: <syntaxhighlight lang='rsplus'>
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")


# Expressions with Spacing
== Game ==
# '~' is to add space and '*' is to squish characters together
* [https://kbroman.org/miner_book/?s=09 R Programming with Minecraft]
plot(1:10, xlab= expression(Delta * 'C'))
* [https://cran.r-project.org/web/packages/pixelpuzzle/index.html pixelpuzzle]
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
* [https://www.rostrum.blog/2022/09/24/pixeltrix/ Interactive pixel art in R with {pixeltrix}]
    ylab = expression(phi^{rho + a} * z * w),
* [https://rtaoist.blogspot.com/2021/03/r-shiny-maths-games-for-6-years-old.html Shiny math game]
    main = "Pure Expressions with Spacing")
* [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]


# Expressions with Text
== Music ==
plot(x,y,
* [https://flujoo.github.io/gm/ gm]. Require to install [https://musescore.org/en MuseScore], an open source and free notation software.
    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
== SAS ==
plot(x,y,
[https://github.com/MangoTheCat/sasMap sasMap] Static code analysis for SAS scripts
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
    main = "Substituted Expressions")
</syntaxhighlight>


=== Impose a line to a scatter plot ===
= R packages =
* abline + lsfit # least squares
[[R_packages|R packages]]
: <syntaxhighlight lang='rsplus'>
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
</syntaxhighlight>
* abline + line # robust line fitting
: <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 ===
= Tricks =
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
<syntaxhighlight lang='rsplus'>
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</syntaxhighlight>


=== Different colors of axis labels in barplot ===
== Getting help ==
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://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/


Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
== Better Coder/coding, best practices ==
<syntaxhighlight lang='rsplus'>
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
tN <- table(Ni <- stats::rpois(100, lambda = 5))
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
r <- barplot(tN, col = rainbow(20))
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
axis(1, 1, LETTERS[1], col.axis="red", col="red")
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
* [https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.9169?campaign=woletoc Best practices in statistical computing] Sanchez 2021
</syntaxhighlight>


Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
<syntaxhighlight lang='rsplus'>
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
</syntaxhighlight>


=== Use [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text text()] to draw labels on X/Y-axis including rotation ===
== Getting user's home directory ==
* adj = 1 means top/rigth alignment. The default is to center the text.
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?]
* [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).
{{Pre}}
** par("usr") is determined *after* a plot has been created
# Windows
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
normalizePath("~")   # "C:\\Users\\brb\\Documents"
* https://datascienceplus.com/building-barplots-with-error-bars/
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
<syntaxhighlight lang='rsplus'>
Sys.getenv("HOME")   # "C:/Users/brb/Documents"
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 ===
# Mac
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in
normalizePath("~")  # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # "/Users/brb"


=== Increase/decrease legend font size ===
# Linux
https://stackoverflow.com/a/36842578
normalizePath("~")  # [1] "/home/brb"
<syntaxhighlight lang='rsplus'>
Sys.getenv("R_USER") # [1] ""
plot(rnorm(100))
Sys.getenv("HOME")   # [1] "/home/brb"
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 ===
== tempdir() ==
Use '''lines()'''.  
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.


Example 1
== Distinguish Windows and Linux/Mac, R.Version() ==
<syntaxhighlight lang='rsplus'>
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))
</syntaxhighlight>


Example 2
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
require(survival)
<pre>
n = 10000
get_os <- function(){
beta1 = 2; beta2 = -1
  sysinf <- Sys.info()
lambdaT = 1 # baseline hazard
  if (!is.null(sysinf)){
lambdaC = 2  # hazard of censoring
    os <- sysinf['sysname']
set.seed(1234)
    if (os == 'Darwin')
x1 = rnorm(n,0)
      os <- "osx"
x2 = rnorm(n,0)
   } else { ## mystery machine
# true event time
    os <- .Platform$OS.type
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))  
    if (grepl("^darwin", R.version$os))
C <- rweibull(n, shape=1, scale=lambdaC)    
      os <- "osx"
time = pmin(T,C)
    if (grepl("linux-gnu", R.version$os))
status <- 1*(T <= C)  
      os <- "linux"
status2 <- 1-status
  }
plot(survfit(Surv(time, status2) ~ 1),
  tolower(os)
    ylab="Survival probability",
}
    main = 'Exponential censoring time')
</pre>
xseq <- seq(.1, max(time), length =100)
<pre>
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
names(R.Version())
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
#  [1] "platform"      "arch"          "os"            "system"       
</syntaxhighlight>
#  [5] "status"        "major"          "minor"          "year"         
#  [9] "month"          "day"            "svn rev"        "language"     
# [13] "version.string" "nickname"
getRversion()
# [1] ‘4.3.0’
</pre>


=== Custom scales ===
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
* 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]
* [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]


== Time series ==
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
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
<pre>
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
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].


=== Time series stock price plot ===
=== What is the best place to save Rconsole on Windows platform ===
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
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://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]
* https://timelyportfolio.github.io/rCharts_time_series/history.html


<syntaxhighlight lang='rsplus'>
My preferred settings:
library(quantmod)
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
getSymbols("AAPL")
* Size: 12
getSymbols("IBM") # similar to AAPL
* background: black
getSymbols("CSCO") # much smaller than AAPL, IBM
* normaltext: white
getSymbols("DJI") # Dow Jones, huge
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow
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))
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


=== Timeline plot ===
A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].
https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date


== Circular plot ==
=== How R starts up ===
* http://freakonometrics.hypotheses.org/20667 which uses https://cran.r-project.org/web/packages/circlize/ circlize] package.
https://rstats.wtf/r-startup.html
* https://www.biostars.org/p/17728/
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.


== Word cloud ==
=== startup - Friendly R Startup Configuration ===
* [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://github.com/henrikbengtsson/startup
* [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]


== World map ==
== Saving and loading history automatically: .Rprofile & local() ==
[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)
<ul>
<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)  


== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
local({r <- getOption("repos")
      r["CRAN"] <- "https://cran.rstudio.com"
      options(repos=r)})


=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
.First <- function(){
* http://rich-iannone.github.io/DiagrammeR/
# library(tidyverse)
* https://donlelek.github.io/2015-03-31-dags-with-r/
cat("\nWelcome at", date(), "\n")
}


=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
.Last <- function(){
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
cat("\nGoodbye at ", date(), "\n")
</pre>
<li>https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
<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.
<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)
</ul>
'''Linux''' or '''Mac'''


=== DAGitty (browser-based and R package) ===
In '''~/.profile''' or '''~/.bashrc''' I put:
* http://dagitty.net/
<pre>
* https://cran.r-project.org/web/packages/dagitty/index.html
export R_HISTFILE=~/.Rhistory
</pre>
In '''~/.Rprofile''' I put:
<pre>
if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>


=== dagR ===
'''Windows'''
* https://cran.r-project.org/web/packages/dagR


== Venn Diagram ==
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.
* limma http://www.ats.ucla.edu/stat/r/faq/venn.htm - only black and white?
<pre>
* VennDiagram - input has to be the numbers instead of the original vector?
if (interactive()) {
* 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]
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
<syntaxhighlight lang='rsplus'>
}
# systemPipeR package method
</pre>
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
== Disable "Save workspace image?" prompt when exit R? ==
source("http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R")
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in 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")
== R release versions ==
counts <- list(sapply(OLlist$Venn_List, length)) 
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
vennPlot(counts=counts)                         
</syntaxhighlight>


[[:File:Vennplot.png]]
== getRversion() ==
<pre>
getRversion()
[1] ‘4.3.0’
</pre>


== Bump chart/Metro map ==
== Detect number of running R instances in Windows ==
https://dominikkoch.github.io/Bump-Chart/
* 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.


== Amazing plots ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
See [[Amazing_plot|Amazing plot]].


== Google Analytics ==
Image Name                    PID Session Name        Session#    Mem Usage
=== GAR package ===
============================================================================
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html
Rgui.exe                      1096 Console                    1    44,712 K


== Linear Programming ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/


== Linear Algebra ==
Image Name                    PID Session Name        Session#    Mem Usage
* [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.
Rserve.exe                    6108 Console                    1    381,796 K
</pre>
In R, we can use
<pre>
> 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"


== Amazon Alexa ==
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
</pre>


== R and Singularity ==
== Editor ==
https://www.rstudio.com/rviews/2017/03/29/r-and-singularity/
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs


== Teach kids about R with Minecraft ==
<ul>
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html
<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).
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
<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


== Secure API keys ==
== GUI for Data Analysis ==
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023


== Hide a password: keyring package ==
=== BlueSky Statistics ===
* https://cran.r-project.org/web/packages/keyring/index.html
* https://www.blueskystatistics.com/Default.asp
* [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]
* [https://r4stats.com/articles/software-reviews/bluesky/ A Comparative Review of the BlueSky Statistics GUI for R]


== Vision and image recognition ==
=== Rcmdr ===
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
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.
* [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


== Turn pictures into coloring pages ==
=== Deducer ===
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
http://cran.r-project.org/web/packages/Deducer/index.html


== Numerical optimization ==
=== jamovi ===
* [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]
* https://www.jamovi.org/
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
* [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


= R packages =
== Scope ==
[[R_packages|R packages]]
See
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.


= Tricks =
=== source() ===
* [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.
* [[#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()''')


== Getting help ==
{{Pre}}
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
## foo.R ##
* https://stat.ethz.ch/pipermail/r-help/
cat(ArrayTools, "\n")
* https://stat.ethz.ch/pipermail/r-devel/
## End of foo.R


== Better Coder/coding, best practices ==
# 1. Error
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
predict <- function() {
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
  ArrayTools <- "C:/Program Files" # or through load() function
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
  source("foo.R")                  # or through a function call; foo()
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]
}
predict()  # Object ArrayTools not found


== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
# 2. OK. Make the variable global
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict()
ArrayTools


== Getting user's home directory ==
# 3. OK. Create a global variable
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?]
ArrayTools <- "C:/Program Files"
<syntaxhighlight lang='rsplus'>
predict <- function() {
# Windows
  source("foo.R")
normalizePath("~")  # "C:\\Users\\brb\\Documents"
}
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
predict()
Sys.getenv("HOME")   # "C:/Users/brb/Documents"
</pre>


# Mac
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
normalizePath("~")  # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # "/Users/brb"


# Linux
Example 1.
normalizePath("~")  # [1] "/home/brb"
<pre>
Sys.getenv("R_USER") # [1] ""
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
Sys.getenv("HOME")   # [1] "/home/brb"
> ttt
</syntaxhighlight>
  type JpnTest
 
1   a    999
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
2    b    999
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
3    c    999
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
4    d    999
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
5    e    999
* [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]
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]
> jpntest()
 
  type JpnTest
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
1    a      N5
<pre>
2    b    999
R_LIBS_SITE=F:/R/library
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>
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''.


Note that on Windows OS, R/etc contains
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.
<pre>
$ ls -l /c/Progra~1/r/r-3.2.0/etc
total 142
-rw-r--r--    1  Administ    1043 Jun 20  2013 Rcmd_environ
-rw-r--r--    1  Administ    1924 Mar 17  2010 Rconsole
-rw-r--r--    1  Administ      943 Oct  3  2011 Rdevga
-rw-r--r--    1  Administ      589 May 20  2013 Rprofile.site
-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
-rw-r--r--    1  Administ    30188 Mar 17  2010 rgb.txt
drwxr-xr-x    3  Administ        0 Jun  8 10:30 x64


$ ls /c/Progra~1/r/r-3.2.0/etc/i386
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
Makeconf


$ cat /c/Progra~1/r/r-3.2.0/etc/Rconsole
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
# Optional parameters for the console and the pager
# The system-wide copy is in R_HOME/etc.
# A user copy can be installed in `R_USER'.


## Style
=== New environment ===
# This can be `yes' (for MDI) or `no' (for SDI).
* http://adv-r.had.co.nz/Environments.html.
  MDI = yes
* [https://www.r-bloggers.com/2011/06/environments-in-r/ Environments in R]
# MDI = no
* 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!


# the next two are only relevant for MDI
Run the same function on a bunch of R objects
toolbar = yes
{{Pre}}
statusbar = no
mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(<nowiki>mye[[n]]</nowiki>)
</pre>


## Font.
Just look at the contents of rda file without saving to anywhere (?load)
# Please use only fixed width font.
<pre>
# If font=FixedFont the system fixed font is used; in this case
local({
# points and style are ignored. If font begins with "TT ", only
  load("myfile.rda")
# True Type fonts are searched for.
  ls()
font = TT Courier New
})
points = 10
</pre>
style = normal # Style can be normal, bold, italic
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.
 
{{Pre}}
# Dimensions (in characters) of the console.
attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
rows = 25
ls(pos = 2)
columns = 80
##  also typically need to cleanup the search path:
# Dimensions (in characters) of the internal pager.
detach("file:all.rda")
pgrows = 25
</pre>
pgcolumns = 80
If we want to read data from internet, '''load()''' works but not attach().
# should options(width=) be set to the console width?
<pre>
setwidthonresize = yes
con <- url("http://some.where.net/R/data/example.rda")
 
## print the value to see what objects were created.
# memory limits for the console scrolling buffer, in chars and lines
print(load(con))
# NB: bufbytes is in bytes for R < 2.7.0, chars thereafter.
close(con)
bufbytes = 250000
# Github example
buflines = 8000
# https://stackoverflow.com/a/62954840
</pre>
[https://stackoverflow.com/a/39621091 source() case].  
<pre>
myEnv <- new.env()   
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))
</pre>


# Initial position of the console (pixels, relative to the workspace for MDI)
=== str( , max) function ===
# xconsole = 0
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]
# yconsole = 0


# Dimension of MDI frame in pixels
If we use str() on a function like str(lm), it is equivalent to args(lm)
# Format (w*h+xorg+yorg) or use -ve w and h for offsets from right bottom
# This will come up maximized if w==0
# MDIsize = 0*0+0+0
# MDIsize = 1000*800+100+0
# MDIsize = -50*-50+50+50  # 50 pixels space all round


# The internal pager can displays help in a single window
For a complicated list object, it is useful to use the '''max.level''' argument; e.g. str(, max.level = 1)
# or in multiple windows (one for each topic)
# pagerstyle can be set to `singlewindow' or `multiplewindows'
pagerstyle = multiplewindows


## Colours for console and pager(s)
For a large data frame, we can use the '''tibble()''' function; e.g. mydf %>% tibble()
# (see rwxxxx/etc/rgb.txt for the known colours).
background = White
normaltext = NavyBlue
usertext = Red
highlight = DarkRed


## Initial position of the graphics window
=== tidy() function ===
## (pixels, <0 values from opposite edge)
broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See [[Tidyverse#broom|here]].
xgraphics = -25
ygraphics = 0


## Language for messages
=== View all objects present in a package, ls() ===
language =
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"


## Default setting for console buffering: 'yes' or 'no'
> ls("package:mypackage")
buffered = yes
[1] "_mypackage_rcpp_hello_world" "evalCpp"                    "library.dynam.unload"     
[4] "rcpp_hello_world"            "system.file"
</pre>
</pre>
and on Linux
<pre>
brb@brb-T3500:~$ whereis R
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


brb@brb-T3500:~$ ls /usr/lib/R
Note that the first argument of ls() (or detach()) is used to specify the environment. It can be
bin  COPYING  etc  lib  library  modules  site-library  SVN-REVISION
* 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).


brb@brb-T3500:~$ ls /usr/lib/R/etc
== Speedup R code ==
javaconf  ldpaths  Makeconf  Renviron  Renviron.orig  Renviron.site  Renviron.ucf  repositories  Rprofile.site
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+


brb@brb-T3500:~$ ls /usr/local/lib/R
=== Profiler ===
site-library
* [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler] (Video)
</pre>
* [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]
and
<pre>
brb@brb-T3500:~$ cat /usr/lib/R/etc/Rprofile.site
##                                              Emacs please make this -*- R -*-
## empty Rprofile.site for R on Debian
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Rprofile and Rprofile.site


# ## Example of .Rprofile
== && vs & ==
# options(width=65, digits=5)
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  
# 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
* The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
# local({
* The longer form evaluates left to right examining only the first element of each vector. The return is one value.
#  # add MASS to the default packages, set a CRAN mirror
* '''The longer form''' evaluates left to right examining only the first element of each vector. '''Evaluation proceeds only until the result is determined.'''
#  old <- getOption("defaultPackages"); r <- getOption("repos")
* 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].
#  r["CRAN"] <- "http://my.local.cran"
* [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.
#  options(defaultPackages = c(old, "MASS"), repos = r)
* [https://www.tjmahr.com/think-of-stricter-logical-operators/ Think of && as a stricter &]
#})
brb@brb-T3500:~$ cat /usr/lib/R/etc/Renviron.site
##                                              Emacs please make this -*- R -*-
## empty Renviron.site for R on Debian
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Renviron and Renviron.site


# ## Example ~/.Renviron on Unix
<pre>
# R_LIBS=~/R/library
c(T,F,T) & c(T,T,T)
# PAGER=/usr/local/bin/less
# [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))


# ## Example .Renviron on Windows
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
# R_LIBS=C:/R/library
# MY_TCLTK="c:/Program Files/Tcl/bin"
 
# ## Example of setting R_DEFAULT_PACKAGES (from R CMD check)
# R_DEFAULT_PACKAGES='utils,grDevices,graphics,stats'
# # this loads the packages in the order given, so they appear on
# # the search path in reverse order.
brb@brb-T3500:~$
</pre>
</pre>


=== What is the best place to save Rconsole on Windows platform ===
== for-loop, control flow ==
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://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Control ?Control]
* '''next''' can be used to skip the rest of the inner-most loop
* [https://www.programiz.com/r/ifelse-function ifelse() Function]


My preferred settings:
== Vectorization ==
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
* Size: 12
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* background: black
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
* normaltext: white
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow
* 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 ===
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]


and others (default options)
=== lapply vs for loop ===
* pagebg: white
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
* pagetext: navy
* https://code-examples.net/en/q/286e03a
* highlight: DarkRed
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


== Saving and loading history automatically: .Rprofile & local() ==
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
* http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html
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?]
* '''.Rprofile''' will automatically be loaded when R has started from that directory
* Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
* 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].
{{Pre}}
* '''.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]].
split(x, f, drop = FALSE, …)
* [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]
</pre>
** You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.  
* [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.
** 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 "+ "
d <- 1:10
options(editor="nano") # default is "vi" on Linux
chunksize <- 4
# options(htmlhelp=TRUE)  
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


local((r <- getOption("repos")
# bigmemory vignette
  r["CRAN"] <- "http://cran.rstudio.com"
planeindices <- split(1:nrow(x), x[,'TailNum'])
  options(repos = r)))
planeStart <- sapply(planeindices,
 
                    function(i) birthmonth(x[i, c('Year','Month'),
.First <- function(){
                                            drop=FALSE]))
# library(Hmisc)
</pre>
cat("\nWelcome at", date(), "\n")
* Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.  
}
{{Pre}}
 
split(mtcars,mtcars$cyl)
.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'''
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
In '''~/.profile''' or '''~/.bashrc''' I put:
<pre>
export R_HISTFILE=~/.Rhistory
</pre>
</pre>
In '''~/.Rprofile''' I put:
* Split columns of a data frame/matrix.  
<pre>
{{Pre}}
if (interactive()) {
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))  
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
# $`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>
</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.


'''Windows'''
sapply(tSsp, function(x) names(which.max(x)))  
 
# return a vector of probset IDs of length of unique entrez IDs
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>
</pre>


== R release versions ==
=== strsplit and sapply ===
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
{{Pre}}
 
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
== Detect number of running R instances in Windows ==
> strsplit(namedf, " ")
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
[[1]]
<pre>
[1] "John" "ABC"
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"
[[2]]
[1] "Mary" "CDE"  


Image Name                    PID Session Name        Session#    Mem Usage
[[3]]
============================================================================
[1] "Kat" "FGH"
Rgui.exe                      1096 Console                    1     44,712 K


C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
> sapply(strsplit(namedf, " "), "[", 1)
 
[1] "John" "Mary" "Kat"
Image Name                    PID Session Name        Session#    Mem Usage
> sapply(strsplit(namedf, " "), "[", 2)
============================================================================
[1] "ABC" "CDE" "FGH"
Rserve.exe                    6108 Console                    1   381,796 K
</pre>
</pre>
In R, we can use
<pre>
> 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
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
</pre>
<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


== Editor ==
# vapply() is safter than sapply().
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
# 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>
</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


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


== GUI for Data Analysis ==
=== Mean of duplicated rows: colMeans and rowsum ===
<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


=== Rcmdr ===
aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
http://cran.r-project.org/web/packages/Rcmdr/index.html
#  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
</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]]


=== Deducer ===
</li>
http://cran.r-project.org/web/packages/Deducer/index.html
<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.''
 
{{Pre}}
=== jamovi ===
group <- rownames(x)
* https://www.jamovi.org/
rowsum(x, group, na.rm=T)/as.vector(table(group))
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
#  [,1] [,2] [,3] [,4] [,5] [,6]
 
# a 10.0 20.0 30.0 40.0 50.0 60.0
== Scope ==
# b  1.5  6.0 11.0 31.5 41.5 51.5
See  
# c  4.0 14.0 24.0 34.0 44.0 54.0
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
# d  7.5 17.5 27.5 37.5 47.5 57.5
* [[#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()''')
</pre>
 
</li>
<syntaxhighlight lang='rsplus'>
</ul>
## foo.R ##
* [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.
cat(ArrayTools, "\n")
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
## End of foo.R
* [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]
<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)


# 1. Error
# Another example: select rows with a minimum value from a certain column (yval in this case)
predict <- function() {
> mydf <- read.table(header=T, text='
  ArrayTools <- "C:/Program Files" # or through load() function
id xval yval
  source("foo.R")                 # or through a function call; foo()
A 1  1
}
A -2  2
predict()  # Object ArrayTools not found
B 3  3
 
B 4  4
# 2. OK. Make the variable global
C 5  5
predict <- function() {
')
  ArrayTools <<- "C:/Program Files'
> x = mydf$xval
  source("foo.R")
> y = mydf$yval
}
> aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
predict() 
   id xval yval
ArrayTools
1  A    1    1
2 B    3    3
3  C    5    5
</pre>
</li>
</ul>


# 3. OK. Create a global variable
=== Mean by Group ===
ArrayTools <- "C:/Program Files"
[https://statisticsglobe.com/mean-by-group-in-r Mean by Group in R (2 Examples) | dplyr Package vs. Base R]
predict <- function() {
<pre>
  source("foo.R")
aggregate(x = iris$Sepal.Length,                # Specify data column
}
          by = list(iris$Species),              # Specify group indicator
predict()
          FUN = mean)                          # Specify function (i.e. mean)
</syntaxhighlight>
</pre>
 
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
 
Example 1.
<pre>
<pre>
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
library(dplyr)
> ttt
iris %>%                                        # Specify data frame
   type JpnTest
  group_by(Species) %>%                        # Specify group indicator
1    a    999
   summarise_at(vars(Sepal.Length),              # Specify column
2    b    999
              list(name = mean))               # Specify function
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>
</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].


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.
== 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 [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


Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
[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?]
* 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.


Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
Some short examples:
* [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?]


=== New environment ===
=== Apply vs for loop ===
http://adv-r.had.co.nz/Environments.html
Note that, apply's performance is not always better than a for loop. See
* 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`?]


Run the same function on a bunch of R objects
=== Progress bar ===
<syntaxhighlight lang='rsplus'>
[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?]
mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(mye[[n]])
</syntaxhighlight>


=== View all objects present in a package, ls() ===
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.
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
<syntaxhighlight lang='rsplus'>
> 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")
[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]
[1] "_mypackage_rcpp_hello_world" "evalCpp"                    "library.dynam.unload"     
[4] "rcpp_hello_world"            "system.file"
</syntaxhighlight>


Note that the first argument of ls() (or detach()) is used to specify the environment. It can be
=== simplify option in sapply() ===
* an integer (the position in the ‘search’ list);
<pre>
* the character string name of an element in the search list;
library(KEGGREST)
* an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).


== Speedup R code ==
names1 <- keggGet(c("hsa05340", "hsa05410"))
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2


== Profiler ==
names3 <- keggGet(c("hsa05340"))
(Video) [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler]
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>


== && vs & ==
=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
<ul>
<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].
<pre>
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)
</pre>
</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.
<pre>
z <- mapply(function(u, v) { max(u, v) },
            u = x[, 1], v = x[, 2])
</pre>
</li>
<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
<ul>
<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().
{{Pre}}
# Syntax: Map(f, ...)


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.
xs <- replicate(5, runif(10), simplify = FALSE)
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


== stopifnot(): function argument sanity check ==
# instead of a more clumsy way
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]
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, ...)


== Vectorization ==
> m1 <- data.frame(id=letters[1:4], val=1:4)
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
> m2 <- data.frame(id=letters[2:6], val=2:6)
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
> merge(m1, m2, "id", all = T)
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
  id val.x val.y
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
1  a    1    NA
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
2  b    2    2
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
3  c    3    3
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
4  d    4    4
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
5  e    NA    5
* [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].
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]
 
=== 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.


=== sapply vs vectorization ===
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].
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]


=== lapply vs for loop ===
=== rapply - recursive version of lapply ===
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
* http://4dpiecharts.com/tag/recursive/
* https://code-examples.net/en/q/286e03a
* [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].
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]


=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
=== replicate ===
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?]
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
* Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
{{Pre}}
:<syntaxhighlight lang='rsplus'>
> replicate(5, rnorm(3))
split(x, f, drop = FALSE, …)
          [,1]      [,2]      [,3]      [,4]        [,5]
</syntaxhighlight>
[1,] 0.2509130 -0.3526600 -0.3170790 1.064816 -0.53708856
* Split rows of a data frame/matrix <syntaxhighlight lang='rsplus'>
[2,0.5222548 1.5343319 0.6120194 -1.811913 -1.09352459
split(mtcars,mtcars$cyl)
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</syntaxhighlight>
</pre>
* 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:_rowMeans|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)))
See [[#parallel_package|parSapply()]] for a parallel version of replicate().
# return a vector of probset IDs of length of unique entrez IDs
</syntaxhighlight>
: And here is another example from the [https://cran.r-project.org/web/packages/bigmemory/vignettes/Overview.pdf bigmemory] vignette,
: <syntaxhighlight lang='rsplus'>
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                    function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
</syntaxhighlight>


=== strsplit and sapply ===
=== Vectorize ===
<syntaxhighlight lang='rsplus'>
* [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()'''.
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
{{Pre}}
> strsplit(namedf, " ")
> 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] "John" "ABC"
[1] 1 1 1 1


[[2]]
[[2]]
[1] "Mary" "CDE"
[1] 2 2 2


[[3]]
[[3]]
[1] "Kat" "FGH"
[1] 3 3


> sapply(strsplit(namedf, " "), "[", 1)
[[4]]
[1] "John" "Mary" "Kat"
[1] 4
> sapply(strsplit(namedf, " "), "[", 2)
</pre>
[1] "ABC" "CDE" "FGH"</syntaxhighlight>
* [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
 
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
</pre>


=== Mean of duplicated columns: rowMeans ===
== plyr and dplyr packages ==
* [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://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
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().
[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
# 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)
[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.
# [1] 31 27 28 29 30 31 32 33 34 35
</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 ===
# plyr has a common syntax -- easier to remember
* colMeans(x, na.rm = FALSE, dims = 1)
# plyr requires less code since it takes care of the input and output format
: <syntaxhighlight lang='rsplus'>
# plyr can easily be run in parallel -- faster
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
rownames(x) <- c(rep("a", 2), rep("b", 3), rep("c", 4), "d")
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
</syntaxhighlight>
* rowsum(x, group, reorder = TRUE, …)
: <syntaxhighlight lang='rsplus'>
x <- matrix(runif(100), ncol = 5) # 20 x 5
group <- sample(1:8, 20, TRUE)
(xsum <- rowsum(x, group)) # 8 x 5
</syntaxhighlight>
* [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)
Tutorials
> mydf <- read.table(header=T, text='
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
id xval yval
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
A 1  1
* [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.
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
</syntaxhighlight>


== Apply family ==
Examples of using dplyr:
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.  
* [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 %>%.  


The following list gives a hierarchical relationship among these functions.
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
* '''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.
** '''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://www.queryhome.com/tech/76799/r-difference-between-apply-vs-sapply-vs-lapply-vs-tapply Difference between apply vs sapply vs lapply vs tapply?]
Tibble objects
* 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
* it does not have row names (cf data frame),
* lapply - When you want to apply a function to each element of a list in turn and get a list back.
* it never changes the type of the inputs (e.g. it never converts strings to factors!),  
* 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.
* it never changes the names of variables
* 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:
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
* [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?]


=== Apply vs for loop ===
{{Pre}}
Note that, apply's performance is not always better than a for loop. See
> data(pew, package = "efficient")
* http://tolstoy.newcastle.edu.au/R/help/06/05/27255.html (answered by Brian Ripley)
> dim(pew)  
* https://stat.ethz.ch/pipermail/r-help/2014-October/422455.html (has one example)
[1] 18 10
* [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 <<-).  ''
> class(pew) # tibble is also a data frame!!
** [http://adv-r.had.co.nz/Functional-programming.html Functional programming]
[1] "tbl_df"    "tbl"        "data.frame"
* [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`?]


=== Progress bar ===
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
[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?]
# 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
</pre>


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


=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
tbObj %>% print(n= nrow(.))
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
</pre>
* 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].
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming
* Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
** 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, ...)


xs <- replicate(5, runif(10), simplify = FALSE)
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.
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


# instead of a more clumsy way
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
lapply(seq_along(xs), function(i) {
{{Pre}}
  weighted.mean(xs[[i]], ws[[i]])
TibbleObject$VarName
})
# OR
</syntaxhighlight>
TibbleObject[["VarName"]]
** 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'>
# OR
# Syntax: Reduce(f, x, ...)
pull(TibbleObject, VarName) # won't be a tibble object anymore


> m1 <- data.frame(id=letters[1:4], val=1:4)
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
> m2 <- data.frame(id=letters[2:6], val=2:6)
# OR
> merge(m1, m2, "id", all = T)
dplyr::select(TibbleObject, 2:5) #
  id val.x val.y
</pre>
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
</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 ===
'''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]
* [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).
my_data <- as_tibble(iris)
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
class(my_data)
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
</pre>


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].
To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)


=== rapply - recursive version of lapply ===
=== llply() ===
* http://4dpiecharts.com/tag/recursive/
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.
* [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].
<pre>
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
</pre>
where rLLID is a list of entrez ID. For example,
<pre>
get("org.Hs.egGO")[["6772"]]
</pre>
returns a list of 49 GOs.


=== replicate ===
=== ddply() ===
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
<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().
=== 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]


=== Vectorize ===
=== Performance/speed comparison ===
* [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'>
[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]
> 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]]
== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
[1] 2 2 2
http://rorynolan.rbind.io/2018/09/30/rcsetseed/


[[3]]
=== get_seed() ===
[1] 3 3
See the same blog
 
{{Pre}}
[[4]]
get_seed <- function() {
[1] 4
  sample.int(.Machine$integer.max, 1)
</syntaxhighlight>
}
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy] <syntaxhighlight lang='rsplus'>
</pre>
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.
[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
=== Random seeds ===
myfunc2(1, 2) # 2
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].
myfunc2(3, 5) # 15
<pre>
myfunc2(c(1,3), c(2,5)) # NA
set.seed(as.numeric(Sys.time()))
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 ==
set.seed(as.numeric(Sys.Date()))  # same seed for each day
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
</pre>


[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
=== .Machine and the largest integer, double ===
 
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].  
[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.
{{Pre}}
 
                          Linux/Mac  32-bit Windows 64-bit Windows
# plyr has a common syntax -- easier to remember
double.eps              2.220446e-16  2.220446e-16  2.220446e-16
# plyr requires less code since it takes care of the input and output format
double.neg.eps          1.110223e-16  1.110223e-16  1.110223e-16
# plyr can easily be run in parallel -- faster
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
 
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
Tutorials
double.base            2.000000e+00  2.000000e+00  2.000000e+00
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
double.digits          5.300000e+01  5.300000e+01  5.300000e+01
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
double.rounding        5.000000e+00  5.000000e+00  5.000000e+00
* [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.
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
</pre>


Examples of using dplyr:
=== NA when overflow ===
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]  
<pre>
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
tmp <- 156287L
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.  
tmp*tmp
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647
</pre>


=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
== How to select a seed for simulation or randomization ==
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
* [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.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]


Tibble objects
== set.seed() allow alphanumeric seeds ==
* it does not have row names (cf data frame),
https://stackoverflow.com/a/10913336
* it never changes the type of the inputs (e.g. it never converts strings to factors!),
* it never changes the names of variables


Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
== set.seed(), for loop and saving random seeds ==
 
<ul>
<syntaxhighlight lang='rsplus'>
<li>[https://www.jottr.org/2020/09/21/detect-when-the-random-number-generator-was-used/ Detect When the Random Number Generator Was Used]
> data(pew, package = "efficient")
<pre>
> dim(pew)  
if (interactive()) {
[1] 18 10
  invisible(addTaskCallback(local({
> class(pew) # tibble is also a data frame!!
    last <- .GlobalEnv$.Random.seed
[1] "tbl_df"    "tbl"        "data.frame"
   
    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>
</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.
<pre>
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!


> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
.Random.seed <- seeds[[23]]  # restore
# A tibble: 162 x 3
data.23 <- runif(5)  
                                                      religion Income Count
data.23
                                                          <chr> <chr> <int>
data[[23]]
1                                                     Agnostic  <$10k    27
</pre>
2                                                      Atheist  <$10k    12
</li>
...
</ul>
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
* [https://www.rdocumentation.org/packages/impute/versions/1.46.0/topics/impute.knn impute.knn]
[1] NA
* 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.''
Warning message:
* 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.''
In mean.default(tidyr::gather(pew, key = Income, value = Count,  :
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
  argument is not numeric or logical: returning NA
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
[1] 181.6975
</syntaxhighlight>


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.
== sample() ==
 
=== sample() inaccurate on very large populations, fixed in R 3.6.0 ===
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17494 The default method for generating from a discrete uniform distribution (used in ‘sample()’, for instance) has been changed]. In prior versions, the probability of generating each integer could vary from equal by up to 0.04% (or possibly more if generating more than a million different integers). See also [https://www.r-bloggers.com/whats-new-in-r-3-6-0/amp/ What's new in R 3.6.0] by David Smith.
<syntaxhighlight lang='rsplus'>
{{Pre}}
TibbleObject$VarName
# R 3.5.3
# OR
set.seed(123)
TibbleObject[["VarName"]]
m <- (2/5)*2^32
# OR
m > 2^31
pull(TibbleObject, VarName) # won't be a tibble object anymore
# [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


dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# R 3.6.0
# OR
# docker run --net=host -it --rm r-base:3.6.0
dplyr::select(TibbleObject, 2:5) #
> set.seed(1234)
</syntaxhighlight>
> sample(5)
 
[1] 4 5 2 3 1
=== llply() ===
> RNGkind(sample.kind = "Rounding")
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.
Warning message:
<pre>
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5
</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() ===
=== Getting different results with set.seed() in RStudio ===
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
[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().''


=== ldply() ===
=== dplyr::sample_n() ===
[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]
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.


== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
== Regular Expression ==
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
See [[Regular_expression|here]].
 
== Read rrd file ==
* https://en.wikipedia.org/wiki/RRDtool
* http://oss.oetiker.ch/rrdtool/
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/


=== get_seed() ===
== on.exit() ==
See the same blog
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.
<syntaxhighlight lang='rsplus'>
<ul>
get_seed <- function() {
<li>Database connections
   sample.int(.Machine$integer.max, 1)
<pre>
library(RSQLite)
sqlite_get_query <- function(db, sql) {
  conn <- dbConnect(RSQLite::SQLite(), db)
  on.exit(dbDisconnect(conn), add = TRUE)
  dbGetQuery(conn, sql)
}
</pre>
<li>File connections
<pre>
read_chars <- function(file_name) {
   conn <- file(file_name, "r")
  on.exit(close(conn), add = TRUE)
  readChar(conn, file.info(file_name)$size)
}
</pre>
<li>Temporary files
<pre>
history_lines <- function() {
  f <- tempfile()
  on.exit(unlink(f), add = TRUE)
  savehistory(f)
  readLines(f, encoding = "UTF-8")
}
</pre>
<li>Printing messages
<pre>
myfun = function(x) {
  on.exit(print("first"))
  on.exit(print("second"), add = TRUE)
  return(x)
}
}
</syntaxhighlight>
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.
</ul>


=== .Machine and the largest integer, double ===
== file, connection ==
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].  
* [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)
<syntaxhighlight lang='rsplus'>
* read() and write()
                          Linux/Mac  32-bit Windows 64-bit Windows
* read.table() and write.table()
double.eps              2.220446e-16   2.220446e-16  2.220446e-16
{{Pre}}
double.neg.eps          1.110223e-16  1.110223e-16   1.110223e-16
out = file('tmp.txt', 'w')
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
writeLines("abcd", out)
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
writeLines("eeeeee", out)
double.base            2.000000e+00   2.000000e+00   2.000000e+00
close(out)
double.digits          5.300000e+01  5.300000e+01  5.300000e+01
readLines('tmp.txt')
double.rounding        5.000000e+00  5.000000e+00  5.000000e+00
unlink('tmp.txt')
double.guard            0.000000e+00  0.000000e+00  0.000000e+00
args(writeLines)
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
 
double.exponent        1.100000e+01  1.100000e+01  1.100000e+01
foo <- function() {
double.min.exp        -1.022000e+03  -1.022000e+03  -1.022000e+03
   con <- file()
double.max.exp          1.024000e+03  1.024000e+03  1.024000e+03
   ...
integer.max            2.147484e+09  2.147484e+09  2.147484e+09
   on.exit(close(con))
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
</pre>
sizeof.pointer          8.000000e+00  4.000000e+00  8.000000e+00
[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
</syntaxhighlight>
<pre>
con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)
</pre>


== How to select a seed for simulation or randomization ==
=== withr package ===
* [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://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].
* [https://www.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]


== set.seed() allow alphanumeric seeds ==
== Clipboard (?connections), textConnection(), pipe() ==
https://stackoverflow.com/a/10913336
<ul>
<li>On Windows, we can use readClipboard() and writeClipboard().
{{Pre}}
source("clipboard")
read.table("clipboard")
</pre></li>
<li>Clipboard -> R. Reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function:
{{Pre}}
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'
</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().
</li>
<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].
* 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
{{Pre}}
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)
</pre>
<li>
<li>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.
</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>
</li>
</ul>


== set.seed(), for loop and saving random seeds ==
=== clipr ===
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.
[https://cran.rstudio.com/web/packages/clipr/ clipr]: Read and Write from the System Clipboard


<pre>
== read/manipulate binary data ==
set.seed(1001)  
* x <- readBin(fn, raw(), file.info(fn)$size)
data <- vector("list", 30)
* rawToChar(x[1:16])
seeds <- vector("list", 30)
* See Biostrings C API
for(i in 1:30) {
 
  seeds[[i]] <- .Random.seed
== String Manipulation ==
  data[[i]] <- runif(5)  
* [https://www.gastonsanchez.com/r4strings/ Handling Strings with R](ebook) by Gaston Sanchez.
}
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
.Random.seed <- seeds[[23]]  # restore
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
data.23 <- runif(5)  
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
data.23
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
data[[23]]
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]
* [https://statisticaloddsandends.wordpress.com/2022/05/05/a-short-note-on-the-startswith-function/ A short note on the startsWith function]
 
=== format(): padding with zero ===
<pre>
ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
[1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
</pre>
 
=== noquote() ===
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.
 
=== 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].
 
=== 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>
library(glue)
name <- "Fred"
glue('My name is {name}.') # My name is Fred.
</pre>
</pre>
* 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.''
</li>
* 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.''
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
</ul>
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].


== sample() inaccurate on very large populations, fixed in R 3.6.0 ==
=== Raw data type ===
* [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'>
[https://twitter.com/hadleywickham/status/1387747735441395712 Fun with strings], [https://en.wikipedia.org/wiki/Cyrillic_alphabets Cyrillic alphabets]
# R 3.5.3
<pre>
set.seed(123)
a1 <- "А"
m <- (2/5)*2^32
a2 <- "A"
m > 2^31
a1 == a2
# [1] FALSE
# [1] FALSE
log10(m)
charToRaw("А")
# [1] 9.23502
# [1] d0 90
x <- sample(m, 1000000, replace = TRUE)
charToRaw("A")
table(x %% 2)
# [1] 41
#     0      1  
</pre>
# 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
=== number of characters limit ===
# docker run --net=host -it --rm r-base:3.6.0
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]
> 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>
== 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


== Regular Expression ==
[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)
See [[Regular_expression|here]].


== Read rrd file ==
== setInternet2 ==
* https://en.wikipedia.org/wiki/RRDtool
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.
* http://oss.oetiker.ch/rrdtool/
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/


== file, connection ==
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.
* [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()
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
* read.table() and write.table()
              "GCF_000001405.13.assembly.txt")
<syntaxhighlight lang='rsplus'>
f1 <- tempfile()
out = file('tmp.txt', 'w')
download.file(url, f1)
writeLines("abcd", out)
</pre>
writeLines("eeeeee", out)
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.
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)
</syntaxhighlight>


== Clipboard (?connections), textConnection(), pipe() ==
The following R command will show the exact svn revision for the R you are currently using.
* On Windows, we can use readClipboard() and writeClipboard(). <syntaxhighlight lang='rsplus'>
<pre>
source("clipboard")
R.Version()$"svn rev"
read.table("clipboard")
</pre>
</syntaxhighlight>
* reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function: <syntaxhighlight lang='rsplus'>
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'
</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 ==
If setInternet2(T), then https protocol is supported in download.file().
* x <- readBin(fn, raw(), file.info(fn)$size)
* rawToChar(x[1:16])
* See Biostrings C API


== String Manipulation ==
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.  
* [http://gastonsanchez.com/blog/resources/how-to/2013/09/22/Handling-and-Processing-Strings-in-R.html ebook] by Gaston Sanchez.
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* 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]
 
=== format(): padding with zero ===
<pre>
ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
</pre>
 
== 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
 
[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)
 
== 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 [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.
<pre>
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "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>
 
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 [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].
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].
Line 3,652: Line 3,864:
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.
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.


== File operation ==
== Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan() ==
* list.files()
In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, '''Inf''' and '''NaN''' as input or output. 
 
See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].
 
[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]
 
== replace() function ==
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)
* https://stackoverflow.com/a/11811147
 
== File/path operations ==
* list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
* file.info()
* file.info()
* dir.create()
* dir.create()
* file.create()
* file.create()
* file.copy()
* 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.


== read/download/source a file from internet ==
== read/download/source a file from internet ==
Line 3,665: Line 3,916:
</pre>
</pre>


=== Zip file and url() function ===
=== Zip, RData, gz file and url() function ===
<pre>
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
</pre>
<pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
Line 3,672: Line 3,926:
</pre>
</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.
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
Another example of using url() is
<pre>
<pre>
load(url("http:/www.example.com/example.RData"))
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].
'''Dropbox''' is easy and works for load(), wget, ...
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.
=== zip function ===
This will include 'hallmarkFiles' root folder in the files inside zip.
<pre>
zip(zipfile = 'myFile.zip',
    files = dir('hallmarkFiles', full.names = TRUE))
# Verify/view the files. 'list = TRUE' won't extract
unzip('testZip.zip', list = TRUE)
</pre>
</pre>


Line 3,682: Line 3,960:


=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
<pre>
{{Pre}}
require(RCurl)
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
Line 3,701: Line 3,979:
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package


== summarytools: create summary tables for vectors and data frames ==
== 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.
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: A frictionless, pipeable approach to dealing with summary statistics ===
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]


== Create publication tables using '''tables''' package ==
=== modelsummary ===
[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
 
=== broom ===
[[Tidyverse#broom|Tidyverse->broom]]
 
=== Create publication tables using '''tables''' package ===
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]


R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,  
R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,  
<syntaxhighlight lang='rsplus'>
{{Pre}}
> library(tables)
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
Line 3,749: Line 4,034:
  Male  mean      4.45  4.31  5.05  
  Male  mean      4.45  4.31  5.05  
         sd        (1.01) (0.93) (0.75)
         sd        (1.01) (0.93) (0.75)
</syntaxhighlight>
</pre>
 
=== 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]


== ClinReport: Statistical Reporting in Clinical Trials ==
=== (archived) ClinReport: Statistical Reporting in Clinical Trials ===
https://cran.r-project.org/web/packages/ClinReport/index.html
https://cran.r-project.org/web/packages/ClinReport/index.html


== Append figures to PDF files ==
== Append figures to PDF files ==
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
== Save base graphics as pseudo-objects ==
[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())
# Display the saved plot
grid::grid.newpage()
p1.base
</pre>


== Extracting tables from PDFs ==  
== Extracting tables from PDFs ==  
* [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'''.  
<ul>
* [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'>
<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'''.  
</li>
<li>
[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.  
{{Pre}}
library(pdftools)
library(pdftools)
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
Line 3,767: Line 4,076:


pdf_data(pdf_file)[[1]]  # data frame/tibble format
pdf_data(pdf_file)[[1]]  # data frame/tibble format
</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.
</pre>  
* [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'>
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
sudo apt install poppler-utils
pdftotext -layout input.pdf output.txt
pdftotext -layout input.pdf output.txt
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</syntaxhighlight>
</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]
 
=== 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) ===
[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.
<pre>
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
</pre>
 
=== stats::xtabs() ===


== Create flat tables in R console using ftable() ==
=== stats::ftable() ===
<syntaxhighlight lang='rsplus'>
{{Pre}}
> ftable(Titanic, row.vars = 1:3)
> ftable(Titanic, row.vars = 1:3)
                   Survived  No Yes
                   Survived  No Yes
Line 3,864: Line 4,238:
     4                  2  6  2  0  0  0
     4                  2  6  2  0  0  0
     5                  0  1  0  0  0  0
     5                  0  1  0  0  0  0
</syntaxhighlight>
</pre>
 
=== [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins] ===
Puts Arbitrary Margins On Multidimensional Tables Or Arrays


== tracemem, data type, copy ==
== tracemem, data type, copy ==
Line 3,917: Line 4,288:
# 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.
# 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].
# 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]


== NA in index ==
== NA in index ==
Line 3,944: Line 4,316:


Don't just use x != "" OR !is.na(x).
Don't just use x != "" OR !is.na(x).
=== Some functions ===
* 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)]


== Constant and 'L' ==
== Constant and 'L' ==
Add 'L' after a constant. For example,
Add 'L' after a constant. For example,
<syntaxhighlight lang='rsplus'>
{{Pre}}
for(i in 1L:n) { }
for(i in 1L:n) { }


Line 3,955: Line 4,331:


n <- length(x);  if(n == 0L) { }
n <- length(x);  if(n == 0L) { }
</syntaxhighlight>
</pre>
 
== Vector/Arrays ==
R indexes arrays from 1 like Fortran, not from 0 like C or Python.
 
=== remove integer(0) ===
[https://stackoverflow.com/a/27980810 How to remove integer(0) from a vector?]
 
=== Append some elements ===
[https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ append() and its after argument]
 
=== 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 ==
== Factor ==
Line 3,961: Line 4,357:
We can specify the factor levels and new labels using the factor() function.
We can specify the factor levels and new labels using the factor() function.


<syntaxhighlight lang='rsplus'>
{{Pre}}
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))
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
</pre>
=== Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when() ===
* [https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html Mastering Data Segmentation: A Guide to Using the cut() Function in R]
:<syntaxhighlight lang='r'>
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
</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


=== Create a factor from a continuous variable: cut() ===
library(tidyverse); library(magrittr)
<syntaxhighlight lang='rsplus'>
set.seed(1)
facVar <- cut(contVar, c(-3, 0, 2, 4), labels = c("low", "medium", "high"))
breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
</syntaxhighlight>
x <- runif(50)
Or [https://youtu.be/7oyiPBjLAWY?t=2480 a tibble object]
bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
<syntaxhighlight lang='rsplus'>
 
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()
</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)
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
Line 3,981: Line 4,423:
           TRUE        ~ "adult")
           TRUE        ~ "adult")
)
)
</syntaxhighlight>
</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>


== Data frame ==
=== How to change one of the level to NA ===
* http://adv-r.had.co.nz/Data-structures.html#data-frames
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/
<pre>
x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA
</pre>


=== stringsAsFactors = FALSE ===
[https://webbedfeet.netlify.app/post/creating-missing-values-in-factors/ Creating missing values in factors]
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/


=== check.names = FALSE ===
=== Concatenating two factor vectors ===
Not trivial. [https://stackoverflow.com/a/5068939 How to concatenate factors, without them being converted to integer level?].
<pre>
<pre>
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
unlist(list(f1, f2))
  1a 2a
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
1  1  1
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1  1  1
2  2  2
</pre>
</pre>


=== Convert data frame factor columns to characters ===
=== droplevels() ===
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
[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.
<syntaxhighlight lang='rsplus'>
 
# Method 1:
=== factor(x , levels = ...) vs levels(x) <-  ===
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
<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.


# Method 2:
{| class="wikitable"
bob[] <- lapply(bob, as.character)
|-
</syntaxhighlight>
| [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
|}


=== data.frame to vector ===
<syntaxhighlight lang='rsplus'>
<syntaxhighlight lang='rsplus'>
> a= matrix(1:6, 2,3)
sizes <- factor(c("small", "large", "large", "small", "medium"))
> rownames(a) <- c("a", "b")
sizes
> colnames(a) <- c("x", "y", "z")
#> [1] small large large small medium
> a
#> Levels: large medium small
  x y z
a 1 3 5
b 2 4 6
> unlist(data.frame(a))
x1 x2 y1 y2 z1 z2
2  3 4 5 6
</syntaxhighlight>


=== merge ===
sizes2 <- factor(sizes, levels = c("small", "medium", "large")) # reorder levels but data is not changed
[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]
sizes2
# [1] small  large  large  small  medium
# Levels: small medium large


=== matrix vs data.frame ===
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'>
<syntaxhighlight lang='rsplus'>
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
set.seed(1)
unique(ip1$Priority)
x <- sample(1:2, 500, replace = TRUE)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
y <- round(x + rnorm(500), 3)
unique(ip1[, "Priority"])  # OK
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 ***


ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
# Correct way if we want to change the baseline level to '2'
unique(ip2$Priority)     # OK
# The estimate was changed by flipping the sign from the original data
sample_data$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>


The length of a matrix and a data frame is different.
=== stats::relevel() ===
<syntaxhighlight lang='rsplus'>
[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.
> length(matrix(1:6, 3, 2))
 
[1] 6
=== reorder(), levels() and boxplot() ===
> length(data.frame(matrix(1:6, 3, 2)))
<ul>
[1] 2
<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)
> x[1]
<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.
  X1
<pre>
1  1
# Syntax:
2  2
# newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
3  3
4  4
5  5
6  6
> x[[1]]
[1] 1 2 3 4 5 6
</syntaxhighlight>
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.


=== Convert a matrix (not data frame) of characters to numeric ===
(bymedian <- with(InsectSprays, reorder(spray, count, median)) )
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
class(bymedian)
<syntaxhighlight lang='rsplus'>
levels(bymedian)
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
boxplot(count ~ bymedian, data = InsectSprays,
    a     b
        xlab = "Type of spray", ylab = "Insect count",
1 0.12 0.567
        main = "InsectSprays data", varwidth = TRUE,
2 0.34 0.890
        col = "lightgray") # boxplots are sorted according to the new levels
> is.data.frame(tmp) # FALSE
boxplot(count ~ spray, data = InsectSprays,
> is.matrix(tmp)     # TRUE
        xlab = "Type of spray", ylab = "Insect count",
> sum(tmp)
        main = "InsectSprays data", varwidth = TRUE,
Error in sum(tmp) : invalid 'type' (character) of argument
        col = "lightgray") # not sorted
> mode(tmp)  # "character"
</pre>
<li>[http://www.deeplytrivial.com/2020/05/statistics-sunday-my-2019-reading.html Statistics Sunday: My 2019 Reading] (reorder function)
</ul>
 
=== factor() vs ordered() ===
<pre>
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


> mode(tmp) <- "numeric"
ordered(levels=c("a", "b", "c"))
> sum(tmp)
# Error in factor(x, ..., ordered = TRUE) :
[1] 1.917
#  argument "x" is missing, with no default
</syntaxhighlight>
</pre>


=== Select Data Frame Columns in R ===
== Data frame ==
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]
* 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.
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/


* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
=== stringsAsFactors = FALSE ===
* select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
* 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


=== Creating data frame using structure() function ===
We can use '''options(stringsAsFactors=FALSE)''' forces R to import character data as character objects.
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]


=== Objects from subsetting a row in a data frame vs matrix ===
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.
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
* [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:
=== check.names = FALSE ===
<syntaxhighlight lang='rsplus'>
Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.
str(trainData)
<pre>
# 'data.frame': 503 obs. of  500 variables:
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
# $ bm001: num 0.429 1 -0.5 1.415 -1.899 ...
  1a 2a
# $ bm002: num 0.0568 1 0.5 0.3556 -1.16 ...
1 1 1
# ...
2 2 2
trainData[1:3, 1:3]
> data.frame("1a"=1:2, "2a"=1:2) # default
#        bm001      bm002    bm003
  X1a X2a
# 1 0.4289449 0.05676296 1.657966
1   1  1
# 2 1.0000000 1.00000000 1.000000
2   2  2
# 3 -0.5000000 0.50000000 0.500000
</pre>
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
 
# Warning message:
=== Create unique rownames: make.unique() ===
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, :
<pre>
#   row names were found from a short variable and have been discarded
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
</syntaxhighlight>
rownames(mydf) <- make.unique(groupCodes)
</pre>
 
=== data.frame() will change rownames ===
<pre>
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"
</pre>


'trees' data from the 'datasets' package
=== Print a data frame without rownames ===
<pre>
<pre>
trees[1:3,]
# Method 1.  
#  Girth Height Volume
rownames(df1) <- NULL
# 1   8.3    70  10.3
# 2  8.6    65  10.3
# 3  8.8    63  10.2


# Wrong ways:
# Method 2.  
data.frame(trees[1,] , trees[2,])
print(df1, row.names = FALSE)
#  Girth Height Volume Girth.1 Height.1 Volume.1
</pre>
# 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:
=== Convert data frame factor columns to characters ===
# method 1: dropping row names
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,])))
{{Pre}}
# OR
# Method 1:
data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
#  time status
 
# 1 8.3    8.6
# Method 2:
# 2 70.0  65.0
bob[] <- lapply(bob, as.character)
# 3 10.3  10.3
</pre>
# 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


is.matrix(trees)
[https://stackoverflow.com/a/2853231 To replace only factor columns]:
# [1] FALSE
<pre>
trees2 <- as.matrix(trees)
# Method 1:
data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
i <- sapply(bob, is.factor)
#        time status
bob[i] <- lapply(bob[i], as.character)
# Girth  8.3    8.6
# Height 70.0  65.0
# Volume 10.3  10.3


dim(trees[1,])
# Method 2:
# [1] 1 3
library(dplyr)
dim(trees2[1, ])
bob %>% mutate_if(is.factor, as.character) -> bob
# 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>
</pre>


== matrix ==
=== Sort Or Order A Data Frame ===
=== NROW vs nrow ===
[https://howtoprogram.xyz/2018/01/07/r-how-to-order-a-data-frame/ How To Sort Or Order A Data Frame In R]
[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.
# 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)


=== matrix (column-major order) multiply a vector ===
=== data.frame to vector ===
* [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) is following the column-major order]
<pre>
df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))


<syntaxhighlight lang='rsplus'>
class(df)
> matrix(1:6, 3,2)
# [1] "data.frame"
    [,1] [,2]
class(t(df))
[1,]    1    4
# [1] "matrix" "array"
[2,]    2    5
class(unlist(df))
[3,]    3    6
# [1] "numeric"
> 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 ==
# Method 1: Convert data frame to matrix using as.matrix()
Use '''unname'''.
# 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)


== format.pval ==
# Method 2: Convert data frame to matrix using t()/transpose
<syntaxhighlight lang='rsplus'>
# and then Convert matrix to vector using as.vector() or c()
> args(format.pval)
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
function (pv, digits = max(1L, getOption("digits") - 2L), eps = .Machine$double.eps,
vec4 <- c(t(df))
    na.form = "NA", ...)  


> format.pval(c(stats::runif(5), pi^-100, NA))
# Not working
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"   
as.vector(df)
> format.pval(c(0.1, 0.0001, 1e-27))
# $x
[1] "1e-01"  "1e-04"  "<2e-16"
# [1] 1 2 3
</syntaxhighlight>
# $y
# [1] 4 5 6


== Customize R: options() ==
# 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>
Q: Why as.vector(df) cannot convert a data frame into a vector?


=== Change the default R repository ===
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.
[[Rstudio#Change_repository|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.
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.


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'''.
=== Using cbind() to merge vectors together? ===
<syntaxhighlight lang='rsplus'>
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.
local({
  r = getOption("repos")
  r["CRAN"] = "https://cran.rstudio.com/"
  options(repos = r)
})
options(continue = "  ")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
  .Last <- function() try(savehistory("~/.Rhistory"))
}
</syntaxhighlight>


=== Change the default web browser ===
=== cbind NULL and data.frame ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey).
[https://9to5tutorial.com/cbind-can-t-combine-null-with-dataframe cbind can't combine NULL with dataframe]. Add as.matrix() will fix the problem.
<syntaxhighlight lang='rsplus'>
> help.start()
If the browser launched by 'xdg-open' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
> /usr/bin/xdg-open: 461: /usr/bin/xdg-open: x-www-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: firefox: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: mozilla: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: epiphany: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: konqueror: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: chromium-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: google-chrome: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links2: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: lynx: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: w3m: not found
xdg-open: no method available for opening 'http://127.0.0.1:27919/doc/html/index.html'
</syntaxhighlight>


The solution is to put
=== 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]
 
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>
options(browser='seamonkey')
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>
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:
=== is.matrix: data.frame is not necessarily a matrix ===
<syntaxhighlight lang='rsplus'>
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.  
> 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 ...
</syntaxhighlight>


We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
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
* [https://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
<pre>
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html
X <- data.frame(x=1:2, y=3:4)
</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.  


=== Change the default editor ===
Another example that is a data frame but not a matrix is the built-in object ''cars''; see ?matrix. It is not a vector
On my Linux and mac, the default editor is "vi". To change it to "nano",
<syntaxhighlight lang='rsplus'>
options(editor = "nano")
</syntaxhighlight>


=== Change prompt and remove '+' sign ===
=== Convert a data frame to a matrix: as.matrix() vs data.matrix() ===
See https://stackoverflow.com/a/1448823.
If I have a data frame X which recorded the time of some files.
<syntaxhighlight lang='rsplus'>
options(prompt="R> ", continue=" ")
</syntaxhighlight>


=== digits ===
* is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
* as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
* [https://stackoverflow.com/a/10712012 ?print.default]
* [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.
* [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
* The default digits 7 may be too small. The acceptable range is 1-22. See the following examples


In R,
<syntaxhighlight lang='r'>
<syntaxhighlight lang='rsplus'>
# latex directory contains cache files from knitting an rmarkdown file
> options()$digits # Default
X <- list.files("latex/", full.names = T) %>%
[1] 7
    grep("RData", ., value=T) %>%
> 100000.07 + .04
    file.info() %>% 
[1] 100000.1
    `[`("mtime")
> options(digits = 16)
X %>% is.data.frame() # TRUE
> 100000.07 + .04
X %>% is.matrix() # FALSE
[1] 100000.11
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>
</syntaxhighlight>


In Python,
* 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.
<syntaxhighlight lang='python'>
* 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).
>>> 100000.07 + .04
* See the following example where as.matrix() and data.matrix() return different resuls.
100000.11
<syntaxhighlight lang='r'>
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"
</syntaxhighlight>
</syntaxhighlight>


=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
=== matrix vs data.frame ===
<syntaxhighlight lang='rsplus'>
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
> numer = 29707; denom = 93874
<pre>
> c(numer/denom, numer, denom)  
Browse[2]> names(res2$surv.data.new[[index]])
[1] 3.164561e-01 2.970700e+04 9.387400e+04
NULL
Browse[2]> colnames(res2$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>


# Method 1. Without changing the global option
Case 2:
> format(c(numer/denom, numer, denom), scientific=FALSE)
{{Pre}}
[1] "   0.3164561" "29707.0000000" "93874.0000000"
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


# Method 2. Change the global option
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
> options(scipen=999)
unique(ip2$Priority)    # OK
> numer/denom
</pre>
[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
</syntaxhighlight>


=== Suppress warnings ===
The length of a matrix and a data frame is different.
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}}
<syntaxhighlight lang='rsplus'>
> length(matrix(1:6, 3, 2))
op <- options("warn")
[1] 6
options(warn = -1)
> length(data.frame(matrix(1:6, 3, 2)))
....
[1] 2
options(op)
> x[1]
 
  X1
# OR
1 1
warnLevel <- options()$warn
2  2
options(warn = -1)
3  3
...
4  4
options(warn = warnLevel)
5  5
</syntaxhighlight>
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.


=== Converts warnings into errors ===
=== How to Remove Duplicates ===
options(warn=2)
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]


== sprintf ==
=== Convert a matrix (not data frame) of characters to numeric ===
=== paste, paste0, sprintf ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
[https://www.r-bloggers.com/paste-paste0-and-sprintf/ this post]
{{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"


=== Format number as fixed width, with leading zeros ===
> mode(tmp) <- "numeric"
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
> sum(tmp)
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1
[1] 1.917
</pre>


<syntaxhighlight lang='rsplus'>
=== Convert Data Frame Row to Vector ===
# sprintf()
as.numeric() or '''c()'''
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


# formatC()
=== Convert characters to integers ===
paste("name", formatC(a, width=3, flag="0"), sep="_")
mode(x) <- "integer"
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


# gsub()
=== Non-Standard Evaluation ===
paste0("bm", gsub(" ", "0", format(5:15)))
[https://thomasadventure.blog/posts/understanding-nse-part1/ Understanding Non-Standard Evaluation. Part 1: The Basics]
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
</syntaxhighlight>


=== sprintf does not print ===
=== Select Data Frame Columns in R ===
Use cat() or print() outside sprintf(). sprintf() do not print in a non interactive mode.
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]
<syntaxhighlight lang='rsplus'>
cat(sprintf('%5.2f\t%i\n',1.234, l234))
</syntaxhighlight>


== Creating publication quality graphs in R ==
* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
* 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


== HDF5 : Hierarchical Data Format==
Another way is to the dollar sign '''$''' operator (?"$") to extract rows or column from a data frame.
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.
<pre>
class(USArrests)  # "data.frame"
USArrests$"Assault"
</pre>
Note that for both data frame and matrix objects, we need to use the '''[''' operator to extract columns and/or rows.
<pre>
USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#        Murder Assault
# Alabama  13.2    236
# Alaska    10.0    263
USArrests[c("Murder", "Assault")]  # all rows


* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
tmp <- data(package="datasets")
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
class(tmp$results)  # "matrix" "array"
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
tmp$results[, "Item"]
* 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.
# Same method can be used if rownames are available in a matrix
 
</pre>
<syntaxhighlight lang='rsplus'>
Note for a '''data.table''' object, we can extract columns using the column names without double quotes.
> h5ls(destination_file)
<pre>
  group                          name      otype  dclass          dim
data.table(USArrests)[1:2, list(Murder, Assault)]
0      /                          data  H5I_GROUP                     
</pre>
1 /data                    expression H5I_DATASET INTEGER 35238 x 65429
2     /                          info  H5I_GROUP                     
3  /info                        author H5I_DATASET  STRING            1
4  /info                        contact H5I_DATASET  STRING            1
5  /info                  creation-date H5I_DATASET  STRING            1
6  /info                            lab H5I_DATASET  STRING            1
7  /info                        version H5I_DATASET  STRING            1
8      /                          meta  H5I_GROUP                     
9  /meta          Sample_channel_count H5I_DATASET  STRING        65429
10 /meta    Sample_characteristics_ch1 H5I_DATASET  STRING        65429
11 /meta        Sample_contact_address H5I_DATASET  STRING        65429
12 /meta            Sample_contact_city H5I_DATASET  STRING        65429
13 /meta        Sample_contact_country H5I_DATASET  STRING        65429
14 /meta      Sample_contact_department H5I_DATASET  STRING        65429
15 /meta          Sample_contact_email H5I_DATASET  STRING        65429
16 /meta      Sample_contact_institute H5I_DATASET  STRING        65429
17 /meta      Sample_contact_laboratory H5I_DATASET  STRING        65429
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 ==
[http://www.econometricsbysimulation.com/2016/12/efficiently-saving-and-sharing-data-in-r_46.html Efficiently Saving and Sharing Data in R]


== Write unix format files on Windows and vice versa ==
=== Add columns to a data frame ===
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html
[https://datasciencetut.com/how-to-add-columns-to-a-data-frame-in-r/ How to add columns to a data frame in R]


== with() and within() functions ==
=== Exclude/drop/remove data frame columns ===
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].
* [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>
closePr <- with(mariokart, totalPr - shipPr)
# method 1
head(closePr, 20)
df = subset(mydata, select = -c(x,z) )


mk <- within(mariokart, {
# method 2
            closePr <- totalPr - shipPr
drop <- c("x","z")
    })
df = mydata[,!(names(mydata) %in% drop)]
head(mk) # new column closePr


mk <- mariokart
# method 3: dplyr
aggregate(. ~ wheels + cond, mk, mean)
mydata2 = select(mydata, -a, -x, -y)
# create mean according to each level of (wheels, cond)
mydata2 = select(mydata, -c(a, x, y))
mydata2 = select(mydata, -a:-y)
mydata2 = mydata[,!grepl("^INC",names(mydata))]
</pre>


aggregate(totalPr ~ wheels + cond, mk, mean)
=== Remove Rows from the data frame ===
[https://datasciencetut.com/remove-rows-from-the-data-frame-in-r/ Remove Rows from the data frame in R]


tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)
=== 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>
</pre>


== stem(): stem-and-leaf plot, bar chart on terminals ==
=== Creating data frame using structure() function ===
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]
* 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 ==
=== Create an empty data.frame ===
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
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)


== 15 Questions All R Users Have About Plots ==
# similar to above
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.
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))


# How To Draw An Empty R Plot? plot.new()
# different data type
# How To Set The Axis Labels And Title Of The R Plots?
a <- data.frame(x1 = character(),
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
                x2 = numeric(),
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext()
                x3 = factor(),
# How To Add Or Change The R Plot’s Legend? legend()
                stringsAsFactors = FALSE)
# 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 ==
=== Objects from subsetting a row in a data frame vs matrix ===
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
* [https://statistical-programming.com/jitter-r-function-example/ The jitter R Function] 3 Example Codes (Basic Application & Boxplot Visualization)
<ul>
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
<li>Subsetting creates repeated rows. This will create unexpected rownames.
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]
 
== Scatterplot with the "rug" function ==
<pre>
<pre>
require(stats) # both 'density' and its default method
R> z <- data.frame(x=1:3, y=2:4)
with(faithful, {
R> rownames(z) <- letters[1:3]
    plot(density(eruptions, bw = 0.15))
R> rownames(z)[c(1,1)]
    rug(eruptions)
[1] "a" "a"
     rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
R> rownames(z[c(1,1),])
})
[1] "a"  "a.1"
R> z[c(1,1), ]
     x y
a  1 2
a.1 1 2
</pre>
</li>
<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:
{{Pre}}
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
</pre>
</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.
== Identify/Locate Points in a Scatter Plot ==
[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]


== Draw a single plot with two different y-axes ==
'trees' data from the 'datasets' package
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html
<pre>
trees[1:3,]
#  Girth Height Volume
# 1  8.3    70  10.3
# 2  8.6    65  10.3
# 3  8.8    63  10.2


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


== SVG ==
# Right ways:
=== Embed svg in html ===
# method 1: dropping row names
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html
data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,])))
 
# OR
=== svglite ===
data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
https://blog.rstudio.org/2016/11/14/svglite-1-2-0/
#  time status
 
# 1  8.3    8.6
=== pdf -> svg ===
# 2 70.0  65.0
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
# 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


== read.table ==
# Method 3: convert a data frame to a matrix
=== clipboard ===
is.matrix(trees)
<syntaxhighlight lang="rsplus">
# [1] FALSE
source("clipboard")
trees2 <- as.matrix(trees)
read.table("clipboard")
data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
</syntaxhighlight>
#        time status
# Girth  8.3    8.6
# Height 70.0  65.0
# Volume 10.3  10.3


=== inline text ===
dim(trees[1,])
<syntaxhighlight lang="rsplus">
# [1] 1 3
mydf <- read.table(header=T, text='
dim(trees2[1, ])
  cond yval
# NULL
    A 2
trees[1, ] # notice the row name '1' on the left hand side
     B 2.5
#  Girth Height Volume
    C 1.6
# 1  8.3     70  10.3
')
trees2[1, ]
</syntaxhighlight>
#  Girth Height Volume
#    8.3  70.0  10.3
</pre>
</li>
</ul>


=== http(s) connection ===
=== Convert a list to data frame ===
<syntaxhighlight lang="rsplus">
[https://www.statology.org/convert-list-to-data-frame-r/ How to Convert a List to a Data Frame in R].  
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
<pre>
                          ssl.verifypeer = FALSE)
# method 1
ip <- read.table(textConnection(temp), as.is=TRUE)
data.frame(t(sapply(my_list,c)))
</syntaxhighlight>


=== read only specific columns ===
# method 2
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.
library(dplyr)
<syntaxhighlight lang="rsplus">
bind_rows(my_list) # OR bind_cols(my_list)
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.
# method 3
<syntaxhighlight lang="rsplus">
library(data.table)
library(magrittr)
rbindlist(my_list)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</pre>
</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]
=== tibble and data.table ===
* [[R#tibble | tibble]]
* [[Tidyverse#data.table|data.table]]


== Serialization ==
=== Clean  a dataset ===
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://finnstats.com/index.php/2021/04/04/how-to-clean-the-datasets-in-r/ How to clean the datasets in R]
[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 ==
== matrix ==
See ?socketconnection.


=== Simple example ===
=== Define and subset a matrix ===
from the socketConnection's manual.
* [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]'''.  


Open one R session
<pre>
<pre>
con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
data <- c(2, 4, 7, 5, 10, 1)
writeLines(LETTERS, con1)
A <- matrix(data, ncol = 3)
close(con1)
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>
</pre>


Open another R session (client)
=== Prevent automatic conversion of single column to vector ===
<pre>
use '''drop = FALSE''' such as mat[, 1, drop = FALSE].
con2 <- socketConnection(Sys.info()["nodename"], port = 22131)
 
# as non-blocking, may need to loop for input
=== complete.cases(): remove rows with missing in any column ===
readLines(con2)
It works on a sequence of vectors, matrices and data frames.
while(isIncomplete(con2)) {
 
  Sys.sleep(1)
=== NROW vs nrow ===
  z <- readLines(con2)
[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.
  if(length(z)) print(z)
}
close(con2)
</pre>


=== Use nc in client ===
=== 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.


The client does not have to be the R. We can use telnet, nc, etc. See the post [https://stat.ethz.ch/pipermail/r-sig-hpc/2009-April/000144.html here]. For example, on the client machine, we can issue
{{Pre}}
<pre>
> matrix(1:6, 3,2)
nc localhost 22131   [ENTER]
    [,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>
</pre>
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
* [https://stackoverflow.com/a/20596490 How to divide each row of a matrix by elements of a vector in R]
<pre>
 
nc -v -w 2 localhost -z 22130-22135
=== add a vector to all rows of a matrix ===
</pre>
[https://stackoverflow.com/a/39443126 add a vector to all rows of a matrix]. sweep() or rep() is the best.
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
=== sparse matrix ===
[https://stackoverflow.com/a/10555270 R convert matrix or data frame to sparseMatrix]


=== Use curl command in client ===
To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, '''as.vector()'''.
On the server,
<pre>
con1 <- socketConnection(port = 8080, server = TRUE)
</pre>


On the client,
== Attributes ==
<pre>
* [https://statisticaloddsandends.wordpress.com/2020/10/19/attributes-in-r/ Attributes in R]
curl --trace-ascii debugdump.txt http://localhost:8080/
* [http://adv-r.had.co.nz/Data-structures.html#attributes Data structures] in "Advanced R"
</pre>


Then go to the server,
== Names ==
<pre>
[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()
while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")


close(con1) # return cursor in the client machine
=== Print a vector by suppressing names ===
</pre>
Use '''unname'''. sapply(, , USE.NAMES = FALSE).


=== Use telnet command in client ===
== format.pval/print p-values/format p values ==
On the server,
[https://rdrr.io/r/base/format.pval.html format.pval()]. By default it will show 5 significant digits (getOption("digits")-2).
<pre>
{{Pre}}
con1 <- socketConnection(port = 8080, server = TRUE)
> format.pval(c(stats::runif(5), pi^-100, NA))
</pre>
[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"


On the client,
R> pvalue
<pre>
[1] 0.0004632104
sudo apt-get install telnet
R> print(pvalue, digits =20)
telnet localhost 8080
[1] 0.00046321036188223807528
abcdefg
R> format.pval(pvalue)
hijklmn
[1] "0.00046321"
qestst
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>


Go to the server,
== Customize R: options() ==
<pre>
readLines(con1, 1)
readLines(con1, 1)
readLines(con1, 1)
close(con1) # return cursor in the client machine
</pre>


Some [http://blog.gahooa.com/2009/01/23/basics-of-telnet-and-http/ tutorial] about using telnet on http request. And [http://unixhelp.ed.ac.uk/tables/telnet_commands.html this] is a summary of using telnet.
=== Change the default R repository, my .Rprofile ===
[[Rstudio#Change_repository|Change R repository]]


== Subsetting ==
Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local '''.Rprofile''' settings take precedence.
[http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Subset-assignment Subset assignment of R Language Definition] and [http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Manipulation-of-functions Manipulation of functions].


The result of the command '''x[3:5] <- 13:15''' is as if the following had been executed
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>
`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)
</pre>


=== Avoid Coercing Indices To Doubles ===
Type '''file.edit("~/.Rprofile")'''
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
{{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>


== modelling ==
=== Change the default web browser for utils::browseURL() ===
=== update() ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put
* [https://www.rdocumentation.org/packages/stats/versions/3.6.1/topics/update ?update]
<pre>
* [https://stackoverflow.com/a/5118337 Reusing a Model Built in R]
options(browser='seamonkey')
</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.


=== Extract all variable names in lm(), glm(), ... ===
For one-time only purpose, we can use the ''browser'' option in help.start() function:
all.vars(formula(Model)[-2])
{{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>


=== as.formula(): use a string in formula in lm(), glm(), ... ===
We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
* [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://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
<syntaxhighlight lang='rsplus'>
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html
? as.formula
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</syntaxhighlight>
* [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'>
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")


# Method 1. The 'Call' portion of the model is reported as “formula = f”
=== Change the default editor ===
# our modeling effort,  
On my Linux and mac, the default editor is "vi". To change it to "nano",
# fully parameterized!
{{Pre}}
f <- as.formula(
options(editor = "nano")
  paste(outcome,
</pre>
        paste(variables, collapse = " + "),  
        sep = " ~ "))
print(f)
# mpg ~ cyl + disp + hp + carb


model <- lm(f, data = mtcars)
=== Change prompt and remove '+' sign ===
print(model)
See https://stackoverflow.com/a/1448823.
{{Pre}}
options(prompt="R> ", continue=" ")
</pre>


# Call:
=== digits ===
#  lm(formula = f, data = mtcars)
* [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]
# Coefficients:
* [https://stackoverflow.com/a/10712012 ?print.default]
(Intercept)         cyl        disp          hp        carb 
* [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
#    34.021595    -1.048523    -0.026906    0.009349    -0.926863 
* [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


# Method 2. eval() + bquote() + ".()"
In R,
format(terms(model))  #  or model$terms
{{Pre}}
# [1] "mpg ~ cyl + disp + hp + carb"
> options()$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>


# The new line of code
In Python,
model <- eval(bquote(  lm(.(f), data = mtcars)  ))
{{Pre}}
>>> 100000.07 + .04
100000.11
</pre>


print(model)
=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
# Call:
[https://datasciencetut.com/how-to-turn-off-scientific-notation-in-r/ How to Turn Off Scientific Notation in R?]
#  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
This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.
> eval(bquote(  lm(f, data = mtcars)  ))
{{Pre}}
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)  
[1] 3.164561e-01 2.970700e+04 9.387400e+04


Call:
# Method 1. Without changing the global option
lm(formula = f, data = mtcars)
> format(c(numer/denom, numer, denom), scientific=FALSE)
[1] "    0.3164561" "29707.0000000" "93874.0000000"


Coefficients:
# Method 2. Change the global option
(Intercept)         cyl        disp          hp        carb 
> options(scipen=999)
  34.021595    -1.048523    -0.026906    0.009349    -0.926863
> numer/denom
</syntaxhighlight>
[1] 0.3164561
* [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()).
> c(numer/denom, numer, denom)
[1]    0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]     0.8 29707.0 93874.0
</pre>


=== Aggregating results from linear model ===
=== Suppress warnings: options() and capture.output() ===
https://stats.stackexchange.com/a/6862
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)


== S3 and S4 methods ==
# OR
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
warnLevel <- options()$warn
* Software for Data Analysis: Programming with R by John Chambers
options(warn = -1)
* Programming with Data: A Guide to the S Language  by John Chambers
...
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
options(warn = warnLevel)
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
</pre>
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package]
 
* http://www.cyclismo.org/tutorial/R/s4Classes.html
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/warning suppressWarnings()]
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
<pre>
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
suppressWarnings( foo() )
* [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
foo <- capture.output(
bar <- suppressWarnings(
{print( "hello, world" );
  warning("unwanted" )} ) )  
</pre>


To get the source code of S4 methods, we can use showMethod(), getMethod() and showMethod(). For example
[https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/capture.output capture.output()]
<syntaxhighlight lang='rsplus'>
<pre>
library(qrqc)
str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")
showMethods("gcPlot")
</pre>
getMethod("gcPlot", "FASTQSummary") # get an error
showMethods("gcPlot", "FASTQSummary") # good.
</syntaxhighlight>


* '''Debug a S4 function'''
=== Converts warnings into errors ===
<syntaxhighlight lang='rsplus'>
options(warn=2)
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
Function: nsFilter (package genefilter)
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")
</syntaxhighlight>


* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
=== demo() function ===
<syntaxhighlight lang='rsplus'>
<ul>
library(IRanges)
<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().
ir <- IRanges(start=c(10, 20, 30), width=5)
<pre>
ir
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>


class(ir)
== sprintf ==
## [1] "IRanges"
=== paste, paste0, sprintf ===
## attr(,"package")
[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]
## [1] "IRanges"


getClassDef(class(ir))
=== sep vs collapse in paste() ===
## Class "IRanges" [package "IRanges"]
* 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
## Slots:
<pre>
##                                                                     
R> paste("a", "A", sep=",")
## Name:            start          width          NAMES    elementType
[1] "a,A"
## Class:        integer        integer characterORNULL      character
R> paste("a", "A", sep=",", collapse="-")
##                                     
[1] "a,A"
## Name:  elementMetadata        metadata
R> paste(c("a", "A"), collapse="-")
## Class: DataTableORNULL            list
[1] "a-A"
##
## 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"
</syntaxhighlight>


=== See what methods work on an object ===
R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
see what methods work on an object, e.g. a GRanges object:  
[1] "a,A - b,B - c,C"
<syntaxhighlight lang='rsplus'>methods(class="GRanges")</syntaxhighlight> Or if you have an object, x: <syntaxhighlight lang='rsplus'>methods(class=class(x))</syntaxhighlight>  
R> paste(letters[1:3], collapse = "-")
[1] "a-b-c"
</pre>


=== View S3 function definition: double colon '::' and triple colon ':::' operators ===
=== 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


* pkg::name returns the value of the exported variable name in namespace pkg
{{Pre}}
* pkg:::name returns the value of the internal variable name
# sprintf()
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


<syntaxhighlight lang='rsplus'>
# formatC()
base::"+"
paste("name", formatC(a, width=3, flag="0"), sep="_")
stats:::coef.default
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
</syntaxhighlight>


=== Read the source code (include Fortran/C, S3 and S4 methods) ===
# gsub()
* [https://github.com/jimhester/lookup#readme lookup] package
paste0("bm", gsub(" ", "0", format(5:15)))
* [https://blog.r-hub.io/2019/05/14/read-the-source/ Read the source]
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
</pre>


=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
=== formatC and prettyNum (prettifying numbers) ===
* mcols: Get or set the metadata columns.
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/formatC formatC() & prettyNum()]
* colData: SummarizedExperiment instances from GenomicRanges
* [[R#format.pval|format.pval()]]
* DataFrame: The DataFrame class extends the DataTable virtual class and supports the storage of any type of object (with length and [ methods) as columns.  
<pre>
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"


For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package
R> x <- .000012345
<syntaxhighlight lang='rsplus'>
R> prettyNum(x)
> mcols(ddsNoPrior[genes, ])
[1] "1.2345e-05"
DataFrame with 2 rows and 21 columns
R> x <- .00012345
  baseMean  baseVar  allZero dispGeneEst    dispFit dispersion  dispIter dispOutlier  dispMAP
R> prettyNum(x)
  <numeric> <numeric> <logical>  <numeric>  <numeric>  <numeric> <numeric>  <logical> <numeric>
[1] "0.00012345"
1 163.5750  8904.607    FALSE  0.06263141 0.03862798  0.0577712        7      FALSE 0.0577712
</pre>
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
</syntaxhighlight>


== findInterval() ==
=== Format(x, scientific = TRUE) ===
Related functions are cuts() and split(). See also
Print numeric data in exponential format, so .0001 prints as 1e-4
* [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]


== order(), rank() and sort() ==
== Creating publication quality graphs in R ==
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
<syntaxhighlight lang='rsplus'>
> 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)]
== HDF5 : Hierarchical Data Format==
[1]  1  2  3  4  5  6  7  8  9 10
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.
> sort(x)
[1]  1  2  3  4  5  6  7  8  9 10
</syntaxhighlight>


== do.call ==
* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
* 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.


Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html. help].
== 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]


* Usage
== Write unix format files on Windows and vice versa ==
: <syntaxhighlight lang='rsplus'>
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html
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.
</syntaxhighlight>
* do.call() is similar to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/lapply lapply()] but not the same.
: <syntaxhighlight lang='rsplus'>
> 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
</syntaxhighlight>


* [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.]
== with() and within() functions ==
:<syntaxhighlight lang='rsplus'>
* [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()]
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
* 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].
> length(tmp)
<pre>
[1] 3
closePr <- with(mariokart, totalPr - shipPr)
> tmp[1:4,]
head(closePr, 20)
  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
mk <- within(mariokart, {
[1] 1 1 2 2 3 3 1 1 2 2 3 3
            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)
</pre>
 
== stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals ==
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
* [https://www.dataanalytics.org.uk/tally-plots-in-r/ Tally plots in R]
* https://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package
 
== Plot histograms as lines ==
https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.
<pre>
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)
</pre>


$Var3
== Histogram with density line ==
[1] + + + + + + - - - - - -
<pre>
Levels: + -
hist(x, prob = TRUE)
lines(density(x), col = 4, lwd = 2)
</pre>
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).


$sep
== Graphical Parameters, Axes and Text, Combining Plots ==
[1] ""
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
> do.call("paste", c(tmp, sep = ""))
 
[1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
== 15 Questions All R Users Have About Plots ==
[12] "b3-"
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.
</syntaxhighlight>
* 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.
:<syntaxhighlight lang='rsplus'>
> 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"  "-"
</syntaxhighlight>
* ''environment'' and ''quote'' arguments.
:<syntaxhighlight lang='rsplus'>
> 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))                    
# How To Draw An Empty R Plot? plot.new()
[1] 2
# How To Set The Axis Labels And Title Of The R Plots?
> eval(call("f", quote(A)))             
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()  
[1] 2
# 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].
> eval(call("f", A), envir = env)        
# How To Add Or Change The R Plot’s Legend? legend()
[1] 4
# How To Draw A Grid In Your R Plot? [https://r-charts.com/base-r/grid/ grid()]  
> eval(call("f", quote(A)), envir = env)
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
[1] 100
# How To Adjust The Size Of Points In An R Plot? cex argument
</syntaxhighlight>
# How To Fit A Smooth Curve To Your R Data? loess() and lines()
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
# How To Add Error Bars In An R Plot? arrows()
:<syntaxhighlight lang='rsplus'>
# How To Save A Plot As An Image On Disc
> foo <- function(a=1, b=2, ...) {
# How To Plot Two R Plots Next To Each Other? '''par(mfrow)'''[which means Multiple Figures (use ROW-wise)], '''gridBase''' package, '''lattice''' package
        list(arg=do.call(c, as.list(match.call())[-1]))  
# How To Plot Multiple Lines Or Points? plot(), lines()
  }
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
> foo()
# What Is The Function Of hjust And vjust In ggplot2?
$arg
NULL
> foo(a=1)
$arg
a
1
> foo(a=1, b=2, c=3)
$arg
a b c
1 2 3
</syntaxhighlight>


=== do.call vs lapply ===
== jitter function ==
https://stackoverflow.com/a/10801883
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* 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.
** jitter(, amount) function adds a random variation between -amount/2 and amount/2 to each element in 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).'''
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
* 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.
* [https://www.r-bloggers.com/2023/09/when-to-use-jitter/ When to use Jitter]
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]


:<syntaxhighlight lang='rsplus'>
:[[File:Jitterbox.png|200px]]
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
[1] "numeric"


$Sepal.Width
== Scatterplot with the "rug" function ==
[1] "numeric"
<pre>
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")
})
</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.


$Petal.Length
== Identify/Locate Points in a Scatter Plot ==
[1] "numeric"
<ul>
<li>[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
<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>


$Petal.Width
== Draw a single plot with two different y-axes ==
[1] "numeric"
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html


$Species
== Draw Color Palette ==
[1] "factor"
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/


> x <- lapply(iris, class)
=== Default palette before R 4.0 ===
> do.call(c, x)
palette() # black, red, green3, blue, cyan, magenta, yellow, gray
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species
  "numeric"    "numeric"    "numeric"    "numeric"    "factor"
</syntaxhighlight>


https://stackoverflow.com/a/10801902
<pre>
* '''lapply''' applies a function '''over a list'''. So there will be several function calls.
# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
* '''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
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
[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.
palette(tropical)
: <syntaxhighlight lang='rsplus'>
plot(1:5, 1:5, col=1:5, pch=16, cex=5)
> X <- list(1:3,4:6,7:9)
</pre>
> lapply(X,mean)
[[1]]
[1] 2


[[2]]
=== New palette in R 4.0.0 ===
[1] 5
[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.
<pre>
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")


[[3]]
R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
[1] 8
R> for(id in palette.pals()) {
> do.call(sum, X)
    scales::show_col(palette.colors(palette = id))
[1] 45
    title(id)
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
     readline("Press [enter] to continue")
[1] 45
   }
> do.call(mean, X) # Error
</pre>
> do.call(rbind,X)
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]
     [,1] [,2] [,3]
<pre>
[1,]    1    2   3
palette("ggplot2")
[2,]    4    5    6
palette(palette()[-1]) # Remove 'black'
[3,]   7    8    9
  # OR palette(palette.colors(palette = "ggplot2")[-1] )
> lapply(X,rbind)
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))
[[1]]
    [,1] [,2] [,3]
[1,]    1    2    3


[[2]]
cc <- palette()
    [,1] [,2] [,3]
palette(c(cc,"purple","brown")) # Add two colors
[1,]   4    5    6
</pre>
<pre>
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502
</pre>


[[3]]
=== evoPalette ===
    [,1] [,2] [,3]
[http://gradientdescending.com/evolve-new-colour-palettes-in-r-with-evopalette/ Evolve new colour palettes in R with evoPalette]
[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
</syntaxhighlight>
Below is a good example to show the difference of lapply() and do.call() - [https://stackoverflow.com/a/42734863 Generating Random Strings].
: <syntaxhighlight lang='rsplus'>
> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
[[1]]
[1] "Y" "D" "G" "A"


[[2]]
=== rtist ===
[1] "B" "W" "K" "N"
[https://github.com/tomasokal/rtist?s=09 rtist]: Use the palettes of famous artists in your own visualizations.


> lapply(x, paste0)
== SVG ==
[[1]]
=== Embed svg in html ===
[1] "Y" "D" "G" "A"
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html


[[2]]
=== svglite ===
[1] "B" "W" "K" "N"
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].


> lapply(x, paste0, collapse= '')
=== pdf -> svg ===
[[1]]
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
[1] "YDGA"


[[2]]
=== svg -> png ===
[1] "BWKN"
[https://laustep.github.io/stlahblog/posts/SVG2PNG.html SVG to PNG] using the [https://cran.rstudio.com/web/packages/gyro/index.html gyro] package


> do.call(paste0, x)
== read.table ==
[1] "YB" "DW" "GK" "AN"
=== clipboard ===
</syntaxhighlight>
{{Pre}}
source("clipboard")
read.table("clipboard")
</pre>


=== do.call + rbind + lapply ===
=== inline text ===
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'>
mydf <- read.table(header=T, text='
x <- readLines(textConnection("---CLUSTER 1 ---
  cond yval
  3
    A 2
4
    B 2.5
5
    C 1.6
6
')
---CLUSTER 2 ---
</pre>
9
10
8
11"))


# create a list of where the 'clusters' are
=== http(s) connection ===
clust <- c(grep("CLUSTER", x), length(x) + 1L)
{{Pre}}
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
                          ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
</pre>


# get size of each cluster
=== read only specific columns ===
clustSize <- diff(clust) - 1L
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.
 
{{Pre}}
# get cluster number
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),
clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
#  
system.time(x <- read.delim("Methylation450k.txt",
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
</pre>
 
To know the number of columns, we might want to read the first row first.
{{Pre}}
library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</pre>


result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
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]
    cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
        , Cluster = .cl
        )
    }))


  result
=== check.names = FALSE in read.table() ===
<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"  


    Object Cluster
gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
[1,] "3"    "1"
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
[2,] "4"    "1"
# [1] "hCG_1642354" "IGH@"       "IGHV1-69"    "IGKV1-5"     "IGKV2-24"    "KRTAP13-2"
[3,] "5"   "1"
# [7] "KRTAP19-1"  "KRTAP2-4"   "KRTAP5-9"    "KRTAP6-3"   "Kua-UEV"
[4,] "6"    "1"
</pre>
[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()).
=== setNames() ===
Change the colnames. See an example from [https://www.tidymodels.org/start/models/ tidymodels]


Another example is to reproduce aggregate(). aggregate() = do.call() + by().
=== Testing for valid variable names ===
<syntaxhighlight lang='rsplus'>
[https://www.r-bloggers.com/testing-for-valid-variable-names/ Testing for valid variable names]
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)
</syntaxhighlight>


== How to get examples from help file ==
=== make.names(): Make syntactically valid names out of character vectors ===
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
* [https://stat.ethz.ch/R-manual/R-devel/library/base/html/make.names.html make.names()]
Method 1:
* 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>
<pre>
example(acf, give.lines=TRUE)
make.names("abc-d") # [1] "abc.d"
</pre>
</pre>
Method 2:
 
<pre>
== Serialization ==
Rd <- utils:::.getHelpFile(?acf)
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
tools::Rd2ex(Rd)
[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>
</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.


== "[" and "[[" with the sapply() function ==
=== Simple example ===
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
from the socketConnection's manual.
 
Open one R session
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
close(con1)
</pre>
</pre>
is the same as
 
Open another R session (client)
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
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)
</pre>
</pre>


== Dealing with date ==
=== Use nc in client ===
 
The client does not have to be the R. We can use telnet, nc, etc. See the post [https://stat.ethz.ch/pipermail/r-sig-hpc/2009-April/000144.html here]. For example, on the client machine, we can issue
<pre>
<pre>
d1 = date()
nc localhost 22131  [ENTER]
class(d1) # "character"
</pre>
d2 = Sys.Date()
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.
class(d2) # "Date"
 
If I use the command
<pre>
nc -v -w 2 localhost -z 22130-22135
</pre>
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.


format(d2, "%a %b %d")
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


library(lubridate); ymd("20140108") # "2014-01-08 UTC"
=== Use curl command in client ===
mdy("08/04/2013") # "2013-08-04 UTC"
On the server,
dmy("03-04-2013") # "2013-04-03 UTC"
<pre>
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
con1 <- socketConnection(port = 8080, server = TRUE)
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>
</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.
* [https://cran.r-project.org/web/packages/anytime/index.html anytime] package


== [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] and deparse/substitute ==
On the client,
* [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.
** 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>
<pre>
> deparse(args(lm))
curl --trace-ascii debugdump.txt http://localhost:8080/
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "
</pre>
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                   
[4] "NULL"   


> deparse(args(lm), width=20)
Then go to the server,
[1] "function (formula, data, "        "    subset, weights, "         
<pre>
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, " 
while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")
[5] "    y = FALSE, qr = TRUE, "      "    singular.ok = TRUE, "       
 
[7] "    contrasts = NULL, "          "    offset, ...) "             
close(con1) # return cursor in the client machine
[9] "NULL"
</pre>
</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).
=== Use telnet command in client ===
<syntaxhighlight lang='rsplus'>
On the server,
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3
<pre>
con1 <- socketConnection(port = 8080, server = TRUE)
</pre>


f1(1:3)
On the client,
f2(1:3)
<pre>
f3(1:3)
sudo apt-get install telnet
telnet localhost 8080
abcdefg
hijklmn
qestst
</pre>


# Or
Go to the server,
myfun <- function(f, a) {
<pre>
    eval(parse(text = f))(a)
readLines(con1, 1)
}
readLines(con1, 1)
myfun("f1", 1:3)
readLines(con1, 1)
myfun("f2", 1:3)
close(con1) # return cursor in the client machine
myfun("f3", 1:3)
</pre>


# Or with lapply
Some [http://blog.gahooa.com/2009/01/23/basics-of-telnet-and-http/ tutorial] about using telnet on http request. And [http://unixhelp.ed.ac.uk/tables/telnet_commands.html this] is a summary of using telnet.
method <- c("f1", "f2", "f3")
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method
</syntaxhighlight>


== The ‘…’ argument ==
== Subsetting ==
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
[http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Subset-assignment Subset assignment of R Language Definition] and [http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Manipulation-of-functions Manipulation of functions].


== Lazy evaluation in R functions arguments ==
The result of the command '''x[3:5] <- 13:15''' is as if the following had been executed
* http://adv-r.had.co.nz/Functions.html
<pre>
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)
</pre>


'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  
=== Avoid Coercing Indices To Doubles ===
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]


* Example 1. By default, R function arguments are lazy.
=== Careful on NA value ===
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<pre>
<pre>
f <- function(x) {
R> mydf = data.frame(a=1:3, b=c(NA,5,6))
   999
R> mydf[mydf$b >5, ]
}
    a  b
f(stop("This is an error!"))
NA NA NA
#> [1] 999
3  3  6
R> mydf[which(mydf$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>
</pre>


* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
=== Implicit looping ===
<pre>
<pre>
add <- function(x) {
set.seed(1)
  force(x)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
  function(y) x + y
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
}
sum(i)       # [1] 6
adders2 <- lapply(1:10, add)
x <- 1:10
adders2[[1]](10)
length(x[i]) # [1] 6
#> [1] 11
x[i[1:3]]     # [1] 1  3  4  6  7  9 10
adders2[[10]](10)
length(x[i[1:3]]) # [1] 7
#> [1] 20
</pre>
</pre>


* Example 3. Default arguments are evaluated inside the function.
== modelling ==
<pre>
=== update() ===
f <- function(x = ls()) {
* [https://www.rdocumentation.org/packages/stats/versions/3.6.1/topics/update ?update]
  a <- 1
* [https://stackoverflow.com/a/5118337 Reusing a Model Built in R]
  x
}


# ls() evaluated inside f:
=== Extract all variable names in lm(), glm(), ... ===
f()
all.vars(formula(Model)[-2])
# [1] "a" "x"


# ls() evaluated in global environment:
=== as.formula(): use a string in formula in lm(), glm(), ... ===
f(ls())
* [https://www.r-bloggers.com/2019/08/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula]
# [1] "add"   "adders" "f"  
* [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?]
{{Pre}}
? as.formula
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</pre>
</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]
{{Pre}}
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")


* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
# Method 1. The 'Call' portion of the model is reported as “formula = f”
<pre>
# our modeling effort,
x <- NULL
# fully parameterized!
if (!is.null(x) && x > 0) {
f <- as.formula(
  paste(outcome,
        paste(variables, collapse = " + "),
        sep = " ~ "))
print(f)
# mpg ~ cyl + disp + hp + carb


}
model <- lm(f, data = mtcars)
</pre>
print(model)


== Backtick sign, infix/prefix/postfix operators ==  
# Call:
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].
#  lm(formula = f, data = mtcars)
#
# Coefficients:
(Intercept)         cyl        disp          hp        carb 
#    34.021595    -1.048523    -0.026906    0.009349    -0.926863 


'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
# Method 2. eval() + bquote() + ".()"
<pre>
format(terms(model))  #  or model$terms
1 + 2    # infix
# [1] "mpg ~ cyl + disp + hp + carb"
+ 1 2   # prefix
 
1 2 +   # postfix
# 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
</pre>
</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


== List data type ==
== Replacement function "fun(x) <- a" ==
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
[https://stackoverflow.com/questions/11563154/what-are-replacement-functions-in-r What are Replacement Functions in R?]
<pre>
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
R> xx <- c(1,3,66, 99)
> do.call(mean, args)
R> "cutoff<-" <- function(x, value){
[1] 5.5
    x[x > value] <- Inf
> mean(c(1:10, NA, NA), na.rm = TRUE)
    x
[1] 5.5
}
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>
</pre>
The statement '''fun(x) <- a''' and R will read '''x <- "fun<-"(x,a) '''


== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
== S3 and S4 methods and signature ==
* http://adv-r.had.co.nz/Exceptions-Debugging.html
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
* Software for Data Analysis: Programming with R by John Chambers
<pre>
* Programming with Data: A Guide to the S Language  by John Chambers
out <- try({
* [https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers, 2016
  a <- 1
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
  b <- "x"
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods]
  a + b
* [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
* [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
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* http://adv-r.had.co.nz/S4.html, http://adv-r.had.co.nz/OO-essentials.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]
 
=== Debug an S4 function ===
* '''showMethods('FUNCTION')'''
* '''getMethod('FUNCTION', 'SIGNATURE') ''' 
* '''debug(, signature)'''
{{Pre}}
> args(debug)
function (fun, text = "", condition = NULL, signature = NULL)  
 
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
Function: nsFilter (package genefilter)
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")


elements <- list(1:10, c(-1, 10), c(T, F), letters)
library(DESeq2)
results <- lapply(elements, log)
showMethods("normalizationFactors") # show the object class
is.error <- function(x) inherits(x, "try-error")
                                    # "DESeqDataSet" in this case.
succeeded <- !sapply(results, is.error)
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code
</pre>
</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().
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>
<pre>
tryCatch(expr, ..., finally)
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))


show_condition <- function(code) {
isdebugged("gsva")
  tryCatch(code,
# [1] TRUE
    error = function(c) "error",
undebug(gsva)
    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
</pre>
</pre>
Below is another snippet from available.packages() function,
<syntaxhighlight lang='rsplus>
z <- tryCatch(download.file(....), error = identity)
if (!inherits(z, "error")) STATEMENTS
</syntaxhighlight>


== Using list type ==
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
=== Avoid if-else or switch ===
{{Pre}}
?plot.stepfun.
library(IRanges)
<pre>
ir <- IRanges(start=c(10, 20, 30), width=5)
y0 <- c(1,2,4,3)
ir
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)
class(ir)
op <- par(mfrow = c(2,2))
## [1] "IRanges"
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
## attr(,"package")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
## [1] "IRanges"
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)
getClassDef(class(ir))
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
## Class "IRanges" [package "IRanges"]
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)
##
 
## Slots:
par(op)
##                                                                     
</pre>
## Name:            start          width          NAMES    elementType
[[:File:StepfunExample.svg]]
## Class:        integer        integer characterORNULL      character
 
##                                     
== Open a new Window device ==
## Name:  elementMetadata        metadata
X11() or dev.new()
## Class: DataTableORNULL            list
 
##
== par() ==
## Extends:
?par
## 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"
</pre>


=== text size and font on main, lab & axis ===
=== Check if a function is an S4 method ===
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
'''isS4(foo)'''
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]


Examples:
=== How to access the slots of an S4 object ===
* cex.main=0.9
* @ will let you access the slots of an S4 object.
* cex.lab=0.8
* 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.
* font.lab=2
* [https://kasperdanielhansen.github.io/genbioconductor/html/R_S4.html#slots-and-accessor-functions R - S4 Classes and Methods] Hansen. '''getClass()''' or '''getClassDef()'''.
* cex.axis=0.8
* font.axis=2
* col.axis="grey50"


=== layout ===
=== setReplaceMethod() ===
http://datascienceplus.com/adding-text-to-r-plot/
* [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?]


=== reset the settings ===
=== See what methods work on an object ===
<syntaxhighlight lang='rsplus'>
see what methods work on an object, e.g. a GRanges object:
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)
<pre>
....
methods(class="GRanges")
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
</pre>
</syntaxhighlight>
Or if you have an object, x:
<pre>
methods(class=class(x))
</pre>  


=== mtext (margin text) vs title ===
=== View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere() ===
* https://datascienceplus.com/adding-text-to-r-plot/
?":::"
* https://datascienceplus.com/mastering-r-plot-part-2-axis/


=== mgp (axis label locations) ===
* pkg::name returns the value of the exported variable name in namespace pkg
# 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.
* pkg:::name returns the value of the internal variable name
# 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).


=== pch ===
<pre>
[[:File:R pch.png]]
base::"+"
stats:::coef.default


See [https://www.statmethods.net/advgraphs/parameters.html here].
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")


* Full circle: pch=16
getS3method("t", "test")
</pre>


=== lty (line type) ===
[https://stackoverflow.com/a/19226817 methods() + getAnywhere() functions]
[[:File:R lty.png]]


See [http://www.sthda.com/english/wiki/line-types-in-r-lty here].
=== Read the source code (include Fortran/C, S3 and S4 methods) ===
* [https://github.com/jimhester/lookup#readme lookup] package
* [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.


=== las (label style) ===
=== S3 method is overwritten ===
0: The default, parallel to the axis
For example, the select() method from dplyr is overwritten by [https://github.com/cran/grpreg/blob/master/NAMESPACE grpreg] package.


1: Always horizontal
An easy solution is to load grpreg before loading dplyr.


2: Perpendicular to the axis
* 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]


3: Always vertical
=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html 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.


=== oma (outer margin), common title for two plots ===
For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package
The following trick is useful when we want to draw multiple plots with a common title.
{{Pre}}
> 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
</pre>


<syntaxhighlight lang='rsplus'>
== Pipe ==
par(mfrow=c(1,2),oma = c(0, 0, 2, 0)) # oma=c(0, 0, 0, 0) by default
* [https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
plot(1:10,  main="Plot 1")
* [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]
plot(1:100,  main="Plot 2")
<ul>
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
<li> a(b(x)) vs '''x |> b() |> a()'''. See [https://twitter.com/henrikbengtsson/status/1335328090390597632 this tweet] in R-dev 2020-12-04.
</syntaxhighlight>
<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>


[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
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>


== Non-standard fonts in postscript and pdf graphics ==
== findInterval() ==
https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41
Related functions are cuts() and split(). See also
* [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]


== 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].


== NULL, NA, NaN, Inf ==
: [[File:R162.png|200px]]
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
 
== 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.


== save() vs saveRDS() ==
== order(), rank() and sort() ==
# saveRDS() can only save one R object while save() does not have this constraint.
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
# 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>
<pre>
x <- 5
> x = sample(10)
saveRDS(x, "myfile.rds")
> x
x2 <- readRDS("myfile.rds")
[1]  4  3 10  7  5 8  6  1  9  2
identical(mod, mod2, ignore.environment = TRUE)
> 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
</pre>
</pre>


== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
=== OS-dependent results on sorting string vector ===
* ==: exact match
Gene symbol case.
* all.equal: compare R objects x and y testing ‘near equality’
<pre>
* identical: The safe and reliable way to test two objects for being exactly equal.
# mac:
<syntaxhighlight lang='rsplus'>
order(c("DC-UbP", "DC2")) # c(1,2)
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.
# linux:  
order(c("DC-UbP", "DC2")) # c(2,1)
</pre>


== testhat ==
Affymetric id case.
* https://github.com/r-lib/testthat
<pre>
* [http://www.win-vector.com/blog/2019/03/unit-tests-in-r/ Unit Tests in R]
# mac:
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"


== tinytest ==
# linux
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
order(c("202800_at", "2028_s_at")) # [1] 1 2
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"
</pre>
It does not matter if we include factor() on the character vector.


[https://cran.r-project.org/web/packages/ttdo/index.html ttdo] adds support of the 'diffobj' package for 'diff'-style comparison of R objects.
The difference is related to locale. See


== Numerical Pitfall ==
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/locales ?locales] in R
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
* On OS, type '''locale'''
<syntaxhighlight lang='bash'>
* [https://stackoverflow.com/questions/39171613/sort-produces-different-results-in-ubuntu-and-windows sort() produces different results in Ubuntu and Windows]
.1 - .3/3
* 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).
## [1] 0.00000000000000001388
<pre>
</syntaxhighlight>
# 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


== Sys.getpid() ==
# Or setting the locale to "C"
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.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>


== How to write R codes ==
=== unique() ===
* [https://youtu.be/7oyiPBjLAWY Code smells and feels] from R Consortium
It seems it does not sort. [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/unique ?unique].
** write simple conditions,
<pre>
** handle class properly,
# mac & linux
** return and exit early,
R> unique(c("DC-UbP", "DC2"))
** polymorphism,
[1] "DC-UbP" "DC2"
** switch(),
</pre>
** case_when(),
** %||%.


== How to debug an R code ==
== do.call ==
[[Debug#R|Debug R]]
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.


== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
[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]
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264


== Path length in dir.create() (PR#17206) ==
Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html help].
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() ==
* Usage
* 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>
do.call(what, args, quote = FALSE, envir = parent.frame())
R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
# what: either a function or a non-empty character string naming the function to be called.
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# quote: a logical value indicating whether to quote the arguments.
# envir: an environment within which to evaluate the call. This will be most useful
#        if what is a character string and the arguments are symbols or quoted expressions.
</pre>
</pre>
* https://stackoverflow.com/questions/44873972/default-r-personal-library-location-is-null. Modify '''$HOME/.Renviron''' by adding a line
* 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.
<pre>
{{Pre}}
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"
> 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>
> 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>
* http://stat.ethz.ch/R-manual/R-devel/library/base/html/libPaths.html. Play with .libPaths()
* [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}}
> 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


On Mac & R 3.4.0 (it's fine)
$Var2
<syntaxhighlight lang='rsplus'>
[1] 1 1 2 2 3 3 1 1 2 2 3 3
> 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)
$Var3
<syntaxhighlight lang='rsplus'>
[1] + + + + + + - - - - - -
> Sys.getenv("R_LIBS_USER")
Levels: + -
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"
</syntaxhighlight>


On Linux & R 3.4.1 (*Problematic*)
$sep
<syntaxhighlight lang='rsplus'>
> Sys.getenv("R_LIBS_USER")
[1] ""
[1] ""
> .libPaths()
> do.call("paste", c(tmp, sep = ""))
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[3] "/usr/lib/R/library"
[12] "b3-"
</syntaxhighlight>
</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?


I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
> eval(call("f", A))                     
<syntaxhighlight lang='rsplus'>
[1] 2
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> eval(call("f", quote(A)))             
> library(devtools)
[1] 2
Error in library(devtools) : there is no package called 'devtools'
> eval(call("f", A), envir = env)       
 
[1] 4
# Specify lib.loc parameter will not help with the dependency package
> eval(call("f", quote(A)), envir = env) 
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
[1] 100
Error: package or namespace load failed for 'devtools':
</pre>
.onLoad failed in loadNamespace() for 'devtools', details:
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
  call: loadNamespace(name)
{{Pre}}
  error: there is no package called 'withr'
> 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>
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>
* 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>
# 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>


# A solution is to redefine .libPaths
=== expand.grid, mapply, vapply ===
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
[https://shikokuchuo.net/posts/10-combinations/ A faster way to generate combinations for mapply and vapply]
> 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].
=== 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.
== Using external data from within another package ==
{{Pre}}
https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
 
      Var1
== How to run R scripts from the command line ==
[1,] "a1+"
https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line
[2,] "b1+"
 
[3,] "a2+"
== How to exit a sourced R script ==
[4,] "b2+"
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script How to exit a sourced R script]
[5,] "a3+"
* [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.'' '''
[6,] "b3+"
 
[7,] "a1-"
== Decimal point & decimal comma ==
[8,] "b1-"
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
[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>
 
=== do.call vs lapply ===
[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())'''


== setting seed locally (not globally) in R ==
* 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://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r
* 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.


== R's internal C API ==
{{Pre}}
https://github.com/hadley/r-internals
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
[1] "numeric"


== cleancall package for C resource cleanup ==
$Sepal.Width
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]
[1] "numeric"


== Random numbers: multivariate normal ==
$Petal.Length
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
[1] "numeric"


The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See
$Petal.Width
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
[1] "numeric"
* 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]
<syntaxhighlight lang='rsplus'>
set.seed(1234)
junk <- biospear::simdata(n=500, p=500, q.main = 10, q.inter = 10,
                          prob.tt = .5, m0=1, alpha.tt= -.5,
                          beta.main= -.5, beta.inter= -.5, b.corr = .7, b.corr.by=25,
                          wei.shape = 1, recr=3, fu=2, timefactor=1)
## Method 1: MASS::mvrnorm()
## This is simdata() has used. It gives different numbers on different OS.
##
library(MASS)
set.seed(1234)
m0 <-1
n <- 500
prob.tt <- .5
p <- 500
b.corr.by <- 25
b.corr <- .7
data <- data.frame(treat = rbinom(n, 1, prob.tt) - 0.5)
n.blocks <- p%/%b.corr.by
covMat <- diag(n.blocks) %x%
  matrix(b.corr^abs(matrix(1:b.corr.by, b.corr.by, b.corr.by, byrow = TRUE) -
                    matrix(1:b.corr.by, b.corr.by, b.corr.by)), b.corr.by, b.corr.by)
diag(covMat) <- 1
data <- cbind(data, mvrnorm(n, rep(0, p), Sigma = covMat))
range(data)
# Mac: -4.963827  4.133723
# Linux/Windows: -4.327635  4.408097
packageVersion("MASS")
# Mac: [1] ‘7.3.49’
# Linux: [1] ‘7.3.49’
# Windows: [1] ‘7.3.47’


R.version$version.string
$Species
# Mac: [1] "R version 3.4.3 (2017-11-30)"
[1] "factor"
# Linux: [1] "R version 3.4.4 (2018-03-15)"
# Windows: [1] "R version 3.4.3 (2017-11-30)"


## Method 2: mvtnorm::rmvnorm()
> x <- lapply(iris, class)
library(mvtnorm)
> do.call(c, x)
set.seed(1234)
Sepal.Length Sepal.Width Petal.Length Petal.Width      Species
sigma <- matrix(c(4,2,2,3), ncol=2)
  "numeric"    "numeric"    "numeric"    "numeric"    "factor"
x <- rmvnorm(n=n, rep(0, p), sigma=covMat)
</pre>
range(x)
# Mac: [1] -4.482566 4.459236
# Linux: [1] -4.482566 4.459236


## Method 3: mvnfast::rmvn()
https://stackoverflow.com/a/10801902
set.seed(1234)
* '''lapply''' applies a function '''over a list'''. So there will be several function calls.
x <- mvnfast::rmvn(n, rep(0, p), covMat)
* '''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.
range(x)
{{Pre}}
# Mac: [1] -4.323585  4.355666
> X <- list(1:3,4:6,7:9)
# Linux: [1] -4.323585  4.355666
> lapply(X,mean)
[[1]]
[1] 2


library(microbenchmark)
[[2]]
library(MASS)
[1] 5
library(mvtnorm)
library(mvnfast)
microbenchmark(v1 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "eigen"),
              v2 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "svd"),
              v3 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "chol"),
              v4 <- rmvn(n, rep(0, p), covMat),
              v5 <- mvrnorm(n, rep(0, p), Sigma = covMat))
Unit: milliseconds
expr      min        lq
v1 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "eigen") 296.55374 300.81089
v2 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "svd") 461.81867 466.98806
v3 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "chol") 118.33759 120.01829
v4 <- rmvn(n, rep(0, p), covMat)  66.64675  69.89383
v5 <- mvrnorm(n, rep(0, p), Sigma = covMat) 291.19826 294.88038
mean    median        uq      max neval  cld
306.72485 301.99339 304.46662 335.6137  100    d
478.58536 470.44085 493.89041 571.7990  100    e
125.85427 121.26185 122.21361 151.1658  100  b 
71.67996  70.52985  70.92923 100.2622  100 a   
301.88144 296.76028 299.50839 346.7049  100  c 
</syntaxhighlight>
A little more investigation shows the eigen values differ a little bit on macOS and Linux.
<syntaxhighlight lang='rsplus'>
set.seed(1234); x <- mvrnorm(n, rep(0, p), Sigma = covMat)
debug(mvrnorm)
# eS --- macOS
# eS2 -- Linux
Browse[2]> range(abs(eS$values - eS2$values))
# [1] 0.000000e+00 1.776357e-15
Browse[2]> var(as.vector(eS$vectors))
[1] 0.002000006
Browse[2]> var(as.vector(eS2$vectors))
[1] 0.001999987
Browse[2]> all.equal(eS$values, eS2$values)
[1] TRUE
Browse[2]> which(eS$values != eS2$values)
  [1]   6  7  8  9  10  11  12  13  14  20  22  23  24  25  26  27  28  29
  ...
[451] 494 495 496 497 499 500
Browse[2]> range(abs(eS$vectors - eS2$vectors))
[1] 0.0000000 0.5636919
</syntaxhighlight>


== rle() running length encoding ==
[[3]]
* https://en.wikipedia.org/wiki/Run-length_encoding
[1] 8
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
> do.call(sum, X)
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
[1] 45
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
* https://rosettacode.org/wiki/Run-length_encoding
[1] 45
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
> do.call(mean, X) # Error
* 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
> do.call(rbind,X)
 
    [,1] [,2] [,3]
== citation() ==
[1,]   1    2    3
<syntaxhighlight lang='rsplus'>
[2,]    4    5    6
citation()
[3,]    7    8    9
citation("MASS")
> lapply(X,rbind)
toBibtex(citation())
[[1]]
</syntaxhighlight>
    [,1] [,2] [,3]
[1,]    1    2    3
 
[[2]]
    [,1] [,2] [,3]
[1,]    4    5    6


== Monitor memory usage ==
[[3]]
* Windows: memory.size(max=TRUE)
    [,1] [,2] [,3]
* Linux
[1,]    7    8    9
** 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.
> mapply(mean, X, trim=c(0,0.5,0.1))
** 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`'''
[1] 2 5 8
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
> mapply(mean, X)  
[1] 2 5 8
</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"


References:
[[2]]
* [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 $PID'' is recommended. It only shows the percentage of memory usage.
[1] "B" "W" "K" "N"
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$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 =
> lapply(x, paste0)
== Books ==
[[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"
</pre>
 
=== do.call + rbind + lapply ===
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"))
 
# 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"
</pre>
 
A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
 
Another example is to reproduce aggregate(). aggregate() = do.call() + by().
{{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>
 
== Run examples ==
When we call help(FUN), it shows the document in the browser. The browser will show
<pre>
example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed
</pre>
 
== How to get examples from help file, example() ==
[https://blog.r-hub.io/2020/01/27/examples/ Code examples in the R package manuals]:
<pre>
# How to run all examples from a man page
example(within)
 
# How to check your examples?
devtools::run_examples()
testthat::test_examples()
</pre>
 
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
Method 1:
<pre>
example(acf, give.lines=TRUE)
</pre>
Method 2:
<pre>
Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)
</pre>
 
== "[" and "[[" with the sapply() function ==
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
</pre>
is the same as
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
</pre>
 
== Dealing with dates ==
* 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")
 
# 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)
</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"
 
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
</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'''.
 
* [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
 
== Nonstandard/non-standard evaluation, deparse/substitute and scoping ==
* [https://www.brodieg.com/2020/05/05/on-nse/ Standard and Non-Standard Evaluation in R]
* [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))
 
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) # 👍
</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"   
 
> 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(…))?]
 
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).
{{Pre}}
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
</pre>
 
=== 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>
 
=== 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]
 
== The ‘…’ argument ==
* 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?]
 
== Functions ==
* https://adv-r.hadley.nz/functions.html
* [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!
 
=== 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.
 
Argument matching is augmented by the functions
* [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].
 
Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
 
=== Check function arguments ===
[https://blog.r-hub.io/2022/03/10/input-checking/ Checking the inputs of your R functions]: '''match.arg()''' , '''stopifnot()'''
 
'''stopifnot()''': function argument sanity check
<ul>
<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>
stopifnot(condition1, condition2, ...)
</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>
 
=== Lazy evaluation in R functions arguments ===
* http://adv-r.had.co.nz/Functions.html
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
* https://twitter.com/_wurli/status/1451459394009550850
 
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.
 
* Example 1. By default, R function arguments are lazy.
<pre>
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
</pre>
 
* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
<pre>
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>
 
* Example 3. Default arguments are evaluated inside the function.
<pre>
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"
</pre>
 
* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
<pre>
x <- NULL
if (!is.null(x) && x > 0) {
 
}
</pre>
 
=== Use of functions as arguments ===
[https://www.njtierney.com/post/2019/09/29/unexpected-function/ Just Quickly: The unexpected use of functions as arguments]
 
=== body() ===
[https://stackoverflow.com/a/51548945 Remove top axis title base plot]
 
=== Return functions in R ===
* [https://win-vector.com/2015/04/03/how-and-why-to-return-functions-in-r/ How and why to return functions in R]
* 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]
 
=== 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.
 
<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>
 
== 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 [http://adv-r.had.co.nz/Functions.html Advanced R] and [https://stackoverflow.com/a/36229703 What do backticks do in R?].
<pre>
iris %>%  `[[`("Species")
</pre>
 
'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
<pre>
1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix
</pre>
 
Use with functions like sapply, e.g. '''sapply(1:5, `+`, 3) '''  .
 
== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
<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)
 
# Method 2:
<pre>
defaultW <- getOption("warn")
options(warn = -1)
[YOUR CODE]
options(warn = defaultW)
</pre>
</li>
<li>try() allows execution to continue even after an error has occurred. You can suppress the message with '''try(..., silent = TRUE)'''.
<pre>
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)
</pre>
</li>
<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().
<pre>
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
</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)
 
== List data type ==
=== Create an empty list ===
<pre>
out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)
 
outlist <- as.list(seq(nfolds))
</pre>
 
=== Using $ in R on a List ===
[https://r-lang.com/dollar-sign-in-r-with-example/ How to Use Dollar sign in R]
 
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
<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>
 
=== Descend recursively through lists ===
<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].
 
=== Avoid if-else or switch ===
?plot.stepfun.
<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]]
 
== Open a new Window device ==
X11() or dev.new()
 
== par() ==
?par
 
=== text size (cex) and font size on main, lab & axis ===
* [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]
 
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.
<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>
 
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://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
<ul>
<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]
 
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].
</li>
</ul>
 
== rle() running length encoding ==
* 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]
* [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]
* https://rosettacode.org/wiki/Run-length_encoding
* 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
 
== citation() ==
{{Pre}}
citation()
citation("MASS")
toBibtex(citation())
</pre>
 
== R not responding request to interrupt stop process ==
[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).
 
== 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 [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().
* [https://cran.r-project.org/web/packages/peakRAM/index.html peakRAM]: Monitor the Total and Peak RAM Used by an Expression or Function
* [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]
 
References:
* [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).
* [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 '''
 
== Monitor Data ==
[https://www.jstatsoft.org/article/view/v098i01?s=09 Monitoring Data in R with the lumberjack Package]
 
== Pushover ==
[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}]
 
[https://cran.r-project.org/web/packages/pushoverr/ pushoverr]
 
= Resource =
== Books ==
* [https://forwards.github.io/rdevguide/ R Development Guide] R Contribution Working Group
* [https://rviews.rstudio.com/2021/11/04/bookdown-org/ An R Community Public Library] 2011-11-04
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
* [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://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://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://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://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
* [https://b-rodrigues.github.io/fput/ 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/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/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/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/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/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/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
* [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
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* 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://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]
* [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://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]
* [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.
<ul>
: <syntaxhighlight lang='rsplus'>
<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.
# R 3.4.1
{{Pre}}
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
# R 3.4.1
setwd("/tmp/efficientR/")
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
setwd("/tmp/efficientR/")
# generated pdf file is located _book/_main.pdf
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
</pre>
</li>
</ul>
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro
* [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
* [https://www.bigbookofr.com/index.html Big Book of R]
* [https://epirhandbook.com/?s=09 R for applied epidemiology and public health]
* [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.
* [https://education.rstudio.com/learn/ RStudio &rarr; Finding Your Way To R]. Beginners/Intermediates/Experts
* [https://deepr.gagolewski.com/index.html Deep R Programming]
 
== Videos ==
[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.
 
=== Webinar ===
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
 
== 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()'''.
 
== License ==
[http://www.win-vector.com/blog/2019/07/some-notes-on-gnu-licenses-in-r-packages/ Some Notes on GNU Licenses in R Packages]
 
[https://moderndata.plot.ly/why-dash-uses-the-mit-license/ Why Dash uses the mit license (and not a copyleft gpl license)]


bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
== Interview questions ==
# generated epub file is located _book/_main.epub.
* Does R store matrices in column-major order or row-major order?
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
** 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.
# 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 ==
* Explain the difference between == and === in R. Provide an example to illustrate their use.
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
** 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.  


== useR! ==
* What is the purpose of the apply() function in R? How does it differ from the for loop?
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
** 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.
* [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]


== R consortium ==
* Describe the concept of factors in R. How are they used in data manipulation and analysis?
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
** 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.


== Blogs, Tips, Socials, Communities ==
* What is the significance
* Google: revolutionanalytics In case you missed it
of the attach() and detach() functions in R? When should they be used?
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
** 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.
* [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 ==
* Explain the concept of vectorization in R. How does it impact the performance of R code?
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.
** 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.


Use '''sessionInfo()'''.
* 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.


== License ==
* 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.
[http://www.win-vector.com/blog/2019/07/some-notes-on-gnu-licenses-in-r-packages/ Some Notes on GNU Licenses in R Packages]
** 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.