R: Difference between revisions

From 太極
Jump to navigation Jump to search
(740 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>


== Bioconductor on Amazon EC2 ==
The Chinese character works too.
http://www.bioconductor.org/help/bioconductor-cloud-ami/
{{Pre}}
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</pre>


= Big Data Analysis =
To read all worksheets we need a convenient function
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
{{Pre}}
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
read_excel_allsheets <- function(filename) {
* [https://rstudio-pubs-static.s3.amazonaws.com/72295_692737b667614d369bd87cb0f51c9a4b.html Handling large data sets in R]
    sheets <- readxl::excel_sheets(filename)
* [https://www.oreilly.com/library/view/big-data-analytics/9781786466457/#toc-start Big Data Analytics with R] by Simon Walkowiak
    sheets <- sheets[-1] # Skip sheet 1
* [https://pbdr.org/publications.html pbdR]
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
** https://en.wikipedia.org/wiki/Programming_with_Big_Data_in_R
    names(x) <- sheets
** [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
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</pre>


== bigmemory, biganalytics, bigtabulate ==
=== [https://cran.r-project.org/web/packages/readr/ readr] ===


== ff, ffbase ==
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.
* 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 ==
[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.


== data.table ==
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.
See [[#data.table_2|data.table]].
* 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!)


= Reproducible Research =
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.
http://cran.r-project.org/web/views/ReproducibleResearch.html
 
Note that '''data.table::fread()''' can read a selection of the columns.
 
=== Speed comparison ===
[https://predictivehacks.com/the-fastest-way-to-read-and-write-file-in-r/ The Fastest Way To Read And Write Files In R]. data.table >> readr >> base.


== Reproducible Environments ==
== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
See [[Ggplot2|ggplot2]]


== Some lessons in R coding ==
== Data Manipulation & Tidyverse ==
# 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 [[Tidyverse|Tidyverse]].
# 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 =
== 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>
 
=== PowerPoint ===
<ul>
<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'''.
<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.
<pre>
svg("svg4.svg", width=4, height=4)
plot(1:10, main="width=4, height=4")
dev.off()


=== Windows 7 ===
svg("svg7.svg", width=7, height=7) # default
To make RInside works on Windows OS, try the following
plot(1:10, main="width=7, height=7")
# 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''.
dev.off()
# 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>
rinside_sample0
</pre>
</pre>
</ul>
=== magick ===
https://cran.r-project.org/web/packages/magick/
See an example [[:File:Progpreg.png|here]] I created.
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().


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}}
library(mvbutils)
foodweb(where = "package:batr")


* RHadoop
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
* Hive
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].
* http://www.techspritz.com/category/tutorials/hadoopmapredcue/ Single node or multinode cluster setup using Ubuntu with VirtualBox (Excellent)
* [http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/ Running Hadoop on Ubuntu Linux (Single-Node Cluster)]
* Ubuntu 12.04 http://www.youtube.com/watch?v=WN2tJk_oL6E and [https://www.dropbox.com/s/05aurcp42asuktp/Chiu%20Hadoop%20Pig%20Install%20Instructions.docx instruction]
* Linux Mint http://blog.hackedexistence.com/installing-hadoop-single-node-on-linux-mint
* http://www.r-bloggers.com/search/hadoop


=== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ===
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.


=== XML ===
According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
* 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
[[:File:GgplotPalette.svg]]
# 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
=== [http://rpubs.com/gaston/colortools colortools] ===
doc.text = gsub('\n', ' ', doc.text)
Tools that allow users generate color schemes and palettes


# Join all the elements of the character vector into a single
=== [https://github.com/daattali/colourpicker colourpicker] ===
# character string, separated by spaces
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
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.
=== eyedroppeR ===
<syntaxhighlight lang='rsplus'>
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]
> 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
== [https://github.com/kevinushey/rex rex] ==
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
Friendly Regular 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 ===
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''


== RCurl ==
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
<syntaxhighlight lang='bash'>
# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
 
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


No google ID is required
Seems not work
<pre>
<pre>
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
library(formatR)
  arguments imply differing number of rows: 2, 0
tidy_source("Input.R", file = "output.R", width.cutoff=70)
tidy_source("clipboard")
# default width is getOption("width") which is 127 in my case.
</pre>
</pre>


=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
Some issues
'''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.
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
<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:
cat("abcd",
* deb: libssl-dev (Debian, Ubuntu, etc)
    # This is my comment
* rpm: openssl-devel (Fedora, CentOS, RHEL)
    "defg")
* 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>
</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!
will result in
 
<pre>
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).
> tidy_source("clipboard")
 
Error in base::parse(text = code, srcfile = NULL) :  
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.
  3:1: unexpected string constant
 
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
3: "defg"
curl is independent of RCurl package.
  ^
 
</pre>
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
* 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.
* https://www.opencpu.org/posts/curl-release-0-8/
 
<syntaxhighlight lang='rsplus'>
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)
</syntaxhighlight>
 
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
 
== [https://cran.r-project.org/web/packages/remotes/index.html 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:
<syntaxhighlight lang='rsplus'>
# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')
</syntaxhighlight>
 
== DirichletMultinomial ==
On Ubuntu, we do
<pre>
<pre>
sudo apt-get install libgsl0-dev
cat("abcd"
    ,"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>


== Create GUI ==
== styler ==
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
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/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]
 
== [http://cran.r-project.org/web/packages/rjson/index.html rjson] ==
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/


== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ==
== Download papers ==
=== Accessing Bitcoin Data with R ===
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
http://blog.revolutionanalytics.com/2015/11/accessing-bitcoin-data-with-r.html
Search and Download Papers from the bioRxiv Preprint Server (biology)


=== Plot IP on google map ===
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
* http://thebiobucket.blogspot.com/2011/12/some-fun-with-googlevis-plotting-blog.html#more  (RCurl, RJONIO, plyr, googleVis)
Interface to the arXiv API
* 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.
=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
<pre>
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
require(RJSONIO) # fromJSON
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
require(RCurl)  # getURL
* https://ropensci.org/technotes/2018/12/14/pdftools-20/


temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
== [https://github.com/ColinFay/aside aside]: set it aside ==
                          ssl.verifypeer = FALSE)
An RStudio addin to run long R commands aside your current session.
ip <- read.table(textConnection(temp), as.is=TRUE)
 
names(ip) <- "IP"
== Teaching ==
nr = nrow(ip)
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics
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){
== Organize R research project ==
  cat(i, "\n")
* [https://cran.r-project.org/web/views/ReproducibleResearch.html CRAN Task View: Reproducible Research]
  try(
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
  Coords[i, 1:2] <- ip2coordinates(ip$IP[i])[c("longitude", "latitude")]
* [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow]. It suggests the [https://github.com/r-lib/here here] package. Don't use '''setwd()''' and '''rm(list = ls())'''.
  )
** [https://rstats.wtf/safe-paths.html Practice safe paths]. Use projects and the [https://cran.r-project.org/web/packages/here/index.html here] package.
}
** In RStudio, if we try to send a few lines of code and one of the line contains '''setwd()''', it will give a message: ''The working directory was changed to XXX inside a notebook chunk. The working directory will be reset when the chunk is finished running. Use the knitr root.dir option in the setup chunk to change the working directory for notebook chunks.''
** [http://jenrichmond.rbind.io/post/how-to-use-the-here-package/ how to use the `here` package]
# append to log-file:
** No update for the ''here'' package after 2020-12. Consider [https://github.com/r-lib/usethis usethis] package (Automate project and package setup).
logfile <- data.frame(ip, Lat = Coords$Lat, Long = Coords$Lon,
* drake project
                                      LatLong = paste(round(Coords$Lat, 1), round(Coords$Lon, 1), sep = ":"))
** [https://ropensci.org/blog/2018/02/06/drake/ The prequel to the drake R package]
log_gmap <- logfile[!is.na(logfile$Lat), ]
** [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]


require(googleVis) # gvisMap
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
gmap <- gvisMap(log_gmap, "LatLong",
[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]
                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
=== Naming convention ===
[http://jeffreyhorner.tumblr.com/page/3 Jeffrey Horner's note about deploying Rook App].
<ul>
<li>[https://stackoverflow.com/a/1946879 What is your preferred style for naming variables in R?]
* Use of period separator: they can get mixed up in simple method dispatch. However, it is used by base R ([https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/make.names make.names()], read.table(), et al)
* Use of underscores: really annoying for ESS users
* '''camelCase''': Winner
<li>However, the [https://stackoverflow.com/a/13413278 survey] said (no surprises perhaps) that
* '''lowerCamelCase''' was most often used for function names,
* '''period.separated''' names most often used for parameters.
<li>[https://datamanagement.hms.harvard.edu/collect/file-naming-conventions What are file naming conventions?]
<li>[https://www.r-bloggers.com/2014/07/consistent-naming-conventions-in-r/ Consistent naming conventions in R]
<li>http://adv-r.had.co.nz/Style.html
<li>[https://www.r-bloggers.com/2011/07/testing-for-valid-variable-names/ Testing for valid variable names]
<li>R reserved words ?Reserved
* [https://www.datamentor.io/r-programming/reserved-words/ R Reserved Words]
* Among these words, if, else, repeat, while, function, for, '''in''', next and break are used for conditions, loops and user defined functions.
<li>Microarray/RNA-seq data
<pre>
clinicalDesignData  # clnDesignData
geneExpressionData  # gExpData
geneAnnotationData  # gAnnoData


== Map ==
dataClinicalDesign
=== [https://rstudio.github.io/leaflet/ leaflet] ===
dataGeneExpression
* rstudio.github.io/leaflet/#installation-and-use
dataAnnotation
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/
</pre>
<pre>
# Search all variables ending with .Data
ls()[grep("\\.Data$", ls())]
# Search all variables starting with data_
ls()[grep("^data_", ls())]
</pre>
</ul>


=== choroplethr ===
=== Efficient Data Management in R ===
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.
* 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 ===
== Text to speech ==
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
== Speech to text ==
See an example from [[R#RJSONIO|RJSONIO]] above.
https://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]


== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
== Weather data ==
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
* [https://github.com/ropensci/prism prism] package
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== gtrendsR - Google Trends ==
== logR ==
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
https://github.com/jangorecki/logR
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
 
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
== Progress bar ==
* [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://github.com/r-lib/progress#readme
* [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 ==
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'.
[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.


# Initial data downloading
== cron ==
# Update existing data
* [https://github.com/bnosac/cronr cronR]
# Create a batch file
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]


== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
== beepr: Play A Short Sound ==
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".
* 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/


== Tool for connecting Excel with R ==
== utils package ==
* https://bert-toolkit.com/
https://www.rdocumentation.org/packages/utils/versions/3.6.2
* [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


== Read/Write Excel files package ==
== tools package ==
* http://www.milanor.net/blog/?p=779
* https://www.rdocumentation.org/packages/tools/versions/3.6.2
* [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.
* [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>)]
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
** It can't be used to open .xls or .xlm files.
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
** [https://fabiomarroni.wordpress.com/2018/08/07/use-r-to-write-multiple-tables-to-a-single-excel-file/ Use R to write multiple tables to a single Excel file]
* [https://github.com/hadley/readxl readxl]: it does not depend on anything although it can only read but not write Excel files. 
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
** [https://github.com/rstudio/webinars/tree/master/36-readxl readxl webinar].
** One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a '''tbl_df''', a '''tbl''' and a '''data.frame.'''
** For writing to Excel formats, use writexl or openxlsx package.
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R


<syntaxhighlight lang='rsplus'>
= Different ways of using R =
library(readxl)
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE,
 
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf,
== 10 things R can do that might surprise you ==
    guess_max = min(1000, n_max), progress = readxl_progress(),
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
    .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).
== R call C/C++ ==
<syntaxhighlight lang='rsplus'>
Mainly talks about .C() and .Call().
> 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.
Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.
<syntaxhighlight lang='rsplus'>
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</syntaxhighlight>


To read all worksheets we need a convenient function
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
<syntaxhighlight lang='rsplus'>
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
read_excel_allsheets <- function(filename) {
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
    sheets <- readxl::excel_sheets(filename)
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
    sheets <- sheets[-1] # Skip sheet 1
* http://www.stat.harvard.edu/ccr2005/
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
* http://mazamascience.com/WorkingWithData/?p=1099
    names(x) <- sheets
* [https://youtube.com/playlist?list=PLwc48KSH3D1OkObQ22NHbFwEzof2CguJJ Make an R package with C++ code] (a playlist from youtube)
    x
* [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
dcfile <- "table0.77_dC_biospear.xlsx"
 
dc <- read_excel_allsheets(dcfile)
=== .Call ===
# Each component (eg dc[[1]]) is a tibble.
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/CallExternal ?.Call]
</syntaxhighlight>
* [http://mazamascience.com/WorkingWithData/?p=1099 Using R — .Call(“hello”)]
* http://adv-r.had.co.nz/C-interface.html
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-callhello/ Using R – .Call(“hello”)]


== [https://cran.r-project.org/web/packages/readr/ readr] ==
Be sure to add the ''PACKAGE'' parameter to avoid an error like
<pre>
cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group,
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) :  
  "standardize" not resolved from current namespace (grpreg)
</pre>


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


[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released.
(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
{{Pre}}
library.dynam("libname", package, lib.loc)
</pre>


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.
=== library.dynam.unload() ===
* 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.
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
* readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
* 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.
* '''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!)
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]


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.
=== gcc ===
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]


Note that '''data.table::fread()''' can read a selection of the columns.
=== Primitive functions ===
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]


== Colors ==
== SEXP ==
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
Some examples from packages
* [http://colorspace.r-forge.r-project.org/articles/colorspace.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes.
* [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.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.
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function


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


[[:File:GgplotPalette.svg]]
== Embedding R ==


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


== Data Manipulation & Tidyverse ==
=== An very simple example (do not return from shell) from Writing R Extensions manual ===
See [[Tidyverse|Tidyverse]].
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>.


== Data Science ==
This example can be run by
See [[Data_science|Data science]] page
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>


== microbenchmark & rbenchmark ==
Note:
* [https://cran.r-project.org/web/packages/microbenchmark/index.html microbenchmark]
# '''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.
** [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]
# '''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''.
* [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] ==
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
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].
=== An example from Bioconductor workshop ===
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c


== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ==
Example:
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
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


== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ==
nano embed.c
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
# 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>


For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
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>
<pre>
sudo apt-get install libgtk2.0-dev libcairo2-dev
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>
</pre>


On Windows OS, we may got the error: '''unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' '''. We need to follow the instruction in [http://tolstoy.newcastle.edu.au/R/e6/help/09/05/15613.html here].
Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].


== [http://igraph.org/r/ igraph] ==
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]


== Identifying dependencies of R functions and scripts ==
=== Create a Simple Socket Server in R ===
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].
<syntaxhighlight lang='rsplus'>
library(mvbutils)
foodweb(where = "package:batr")


foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
Create an R function
 
<pre>
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
simpleServer <- function(port=6543)
</syntaxhighlight>
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.


== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
Welcome to R!
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
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                 


Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
R> quit
Connection closed by foreign host.
</pre>


== Colors ==
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
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]].
* 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] ===
See my [[Rserve]] page.
Tools that allow users generate color schemes and palettes


=== [https://github.com/daattali/colourpicker colourpicker] ===
=== outsider ===
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
* [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]


=== [https://cran.r-project.org/web/packages/inlmisc/index.html inlmisc] ===
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
[https://owi.usgs.gov/blog/tolcolors/ GetTolColors()]. Lots of examples.


== [https://github.com/kevinushey/rex rex] ==
=== [http://rdotnet.codeplex.com/ R.NET] ===
Friendly Regular Expressions


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


See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
Terminal
{{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'''


<pre>
Now go back to R
library(formatR)
{{Pre}}
tidy_source("Input.R", file = "output.R", width.cutoff=70)
install.packages("rJava")
tidy_source("clipboard")
# default width is getOption("width") which is 127 in my case.
</pre>
</pre>
Done!


Some issues
If above does not work, a simple way is by (under Ubuntu) running
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
<pre>
<pre>
cat("abcd",
sudo apt-get install r-cran-rjava
    # This is my comment
    "defg")
</pre>
</pre>
will result in
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.
 
=== RCaller ===
 
=== 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>
> tidy_source("clipboard")
$ Rscript --help
Error in base::parse(text = code, srcfile = NULL) :
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
  3:1: unexpected string constant
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
3: "defg"
  ^
</pre>
</pre>
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
 
Example:
<pre>
<pre>
cat("abcd"
args = commandArgs(trailingOnly=TRUE)
    ,"defg"  # This is my comment
# test if there is at least one argument: if not, return an error
   ,"ghij")
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 become
<pre>
<pre>
cat("abcd", "defg" # This is my comment
Rscript --vanilla sillyScript.R iris.txt out.txt
, "ghij")
# args[1] = iris.txt
# args[2] =  out.txt
</pre>
</pre>
Still bad!!
 
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
=== Rscript, #! Shebang and optparse package ===
<ul>
<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("</p>",
#!/usr/bin/env Rscript
"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
print ("shebang works")
ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),
file=ExternalFileName, sep="\n", append=T)
</pre>
</pre>
will result in
Then in the command line
<pre>
<pre>
> tidy_source("clipboard", width.cutoff=70)
chmod u+x shebang.R
Error in base::parse(text = code, srcfile = NULL) :
./shebang.R
  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>
</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.
<li>[http://www.cureffi.org/2014/01/15/running-r-batch-mode-linux/ Running R in batch mode on Linux]
<pre>
<li>[https://cran.r-project.org/web/packages/optparse/index.html optparse] package. Check out its vignette.
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",
<li>[https://cran.r-project.org/web/packages/getopt/index.html getopt]: C-Like 'getopt' Behavior.
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,
</ul>
    sep = "\n", append = T)
 
</pre>
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
* It merges lines though I don't always want to do that. For example
Provides hash-bang (#!) capability for R
<pre>
 
cat("abcd"
FAQs:
    ,"defg" 
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
  ,"ghij")
* [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]
</pre>
* [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?]
will become
{{Pre}}
<pre>
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
cat("abcd", "defg", "ghij")
# 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()"
</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.


=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
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.
Interface to the arXiv API


=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
'''r''' was not meant to run interactively like '''R'''. See ''man r''.
* 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 ==
=== RInside: Embed R in C++ ===
An RStudio addin to run long R commands aside your current session.
See [[R#RInside|RInside]]


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


== Organize R research project ==
The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
* [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())'''.
** 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) ===
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://rcrastinate.rbind.io/post/how-to-save-and-load-data-in-r-an-overview/ How to save (and load) datasets in R: An overview]


== Text to speech ==
To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>


== Weather data ==
The real build process looks like (check <Makefile> for completeness)
* [https://github.com/ropensci/prism prism] package
<pre>
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]
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>


== logR ==
Hello World example of embedding R in C++.
https://github.com/jangorecki/logR
<pre>
#include <RInside.h>                    // for the embedded R via RInside


== Progress bar ==
int main(int argc, char *argv[]) {
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'.
    RInside R(argc, argv);              // create an embedded R instance


== cron ==
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
[https://github.com/bnosac/cronr cronR]


= Different ways of using R =
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)


== 10 things R can do that might surprise you ==
    exit(0);
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
}
</pre>


== R call C/C++ ==
The above can be compared to the Hello world example in Qt.
Mainly talks about .C() and .Call().
<pre>
#include <QApplication.h>
#include <QPushButton.h>


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.
int main( int argc, char **argv )
{
    QApplication app( argc, argv );


* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
    QPushButton hello( "Hello world!", 0 );
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
    hello.resize( 100, 30 );
* 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


=== NAMESPACE file & useDynLib ===
    app.setMainWidget( &hello );
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
    hello.show();
* 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
    return app.exec();
<syntaxhighlight lang='rsplus'>
}
library.dynam("libname", package, lib.loc)
</pre>
</syntaxhighlight>


=== library.dynam.unload() ===
=== [http://www.rfortran.org/ RFortran] ===
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
RFortran is an open source project with the following aim:
* 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 ===
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]


== SEXP ==
It works only on Windows platform with Microsoft Visual Studio installed:(
Some examples from packages


* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
== Call R from other languages ==
=== C ===
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


== R call Fortran ==
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]
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
* [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 ==
Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace


* 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.
=== JRI ===
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
http://www.rforge.net/JRI/
* [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 ===
=== ryp2 ===
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>.
http://rpy.sourceforge.net/rpy2.html


This example can be run by
== Create a standalone Rmath library ==
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
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].


Note:
Here is my experience based on R 3.0.2 on Windows OS.
# '''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''.


More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
=== 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 [[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.
=== An example from Bioconductor workshop ===
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c
 
Example:
Create <embed.c> file
<pre>
<pre>
#include <Rembedded.h>
cd C:\R\R-3.0.2\src\nmath\standalone
#include <Rdefines.h>
make -f Makefile.win
</pre>


static void doSplinesExample();
=== Use Rmath library in our code ===
int
<pre>
main(int argc, char *argv[])
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
{
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
    Rf_initEmbeddedR(argc, argv);
# It is not LD_LIBRARY_PATH in above.
    doSplinesExample();
    Rf_endEmbeddedR(0);
    return 0;
}
static void
doSplinesExample()
{
    SEXP e, result;
    int errorOccurred;


    // create and evaluate 'library(splines)'
# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
    PROTECT(e = lang2(install("library"), mkString("splines")));
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
# It is OK to save the cpp file under any directory.
    if (errorOccurred) {
        // handle error
    }
    UNPROTECT(1);


    // 'options(FALSE)' ...
# Force to link against the static library <libRmath.a>
    PROTECT(e = lang2(install("options"), ScalarLogical(0)));
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
    // ... modified to 'options(example.ask=FALSE)' (this is obscure)
# OR
    SET_TAG(CDR(e), install("example.ask"));
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe
    R_tryEval(e, R_GlobalEnv, NULL);
    UNPROTECT(1);


    // 'example("ns")'
# Force to link against dynamic library <Rmath.dll>
    PROTECT(e = lang2(install("example"), mkString("ns")));
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
    UNPROTECT(1);
}
</pre>
</pre>
Then build the executable. Note that I don't need to create R_HOME variable.
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>
cd
c:\R>RmathEx1
tar xzvf
Enter a argument for the normal cdf:
cd R-3.0.1
1
./configure --enable-R-shlib
Enter a argument for the chi-squared cdf:
make
1
cd tests/Embedding
Prob(Z <= 1) = 0.841345
make
Prob(Chi^2 <= 1)= 0.682689
~/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
</pre>
</pre>


Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
Below is the cpp program <RmathEx1.cpp>.
<pre>
<pre>
export R_HOME=/home/brb/Downloads/R-3.0.2
//RmathEx1.cpp
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
#define MATHLIB_STANDALONE
./embed # No need to include R CMD in front.
#include <iostream>
</pre>
#include "Rmath.h"


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].
using std::cout; using std::cin; using std::endl;


Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
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;


=== Create a Simple Socket Server in R ===
  cout << "Prob(Z <= " << x1 << ") = " <<
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].
     pnorm(x1, 0, 1, 1, 0) << endl;
 
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
Create an R function
     pchisq(x2, 1, 1, 0) << endl;
<pre>
   return 0;
simpleServer <- function(port=6543)
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
     cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
     writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
   }
}
}
</pre>
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.


Welcome to R!
== Calling R.dll directly ==
R> summary(iris[, 3:5])
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.
  Petal.Length    Petal.Width          Species 
 
Min.   :1.000  Min.   :0.100  setosa    :50 
== Create HTML report ==
1st Qu.:1.600  1st Qu.:0.300  versicolor:50 
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].
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
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
Connection closed by foreign host.
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>


=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
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]].
* 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]


See my [[Rserve]] page.
=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
* https://github.com/renkun-ken/formattable
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]


=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
=== [https://github.com/crubba/htmltab htmltab] package ===
This package is NOT used to CREATE html report but EXTRACT html table.


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


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


Terminal
== Create pdf and epub files ==
<syntaxhighlight lang='bash'>
{{Pre}}
# jdk 7
# Idea:
sudo apt-get install openjdk-7-*
#        knitr        pdflatex
update-alternatives --config java
#   rnw -------> tex ----------> pdf
# oracle jdk 8
library(knitr)
sudo add-apt-repository -y ppa:webupd8team/java
knit("example.rnw") # create example.tex file
sudo apt-get update
</pre>
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
{{Pre}}
sudo apt-get -y install openjdk-8-jdk
git clone https://github.com/yihui/knitr-examples.git
</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'''
* <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.


Now go back to R
Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
<syntaxhighlight lang='rsplus'>
{{Pre}}
install.packages("rJava")
# Idea:
</syntaxhighlight>
#        knitr        pandoc
Done!
#  rmd -------> md ----------> pdf


If above does not work, a simple way is by (under Ubuntu) running
git clone https://github.com/yihui/knitr-examples.git
<pre>
cd knitr-examples
sudo apt-get install r-cran-rjava
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</pre>
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


=== RCaller ===
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


=== RApache ===
library(knitr)
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf
knit("DESeq2.Rnw") # create DESeq2.tex
 
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
</pre>
Provides hash-bang (#!) capability for R


FAQs:
Convert tex to epub
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
<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
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
                                              # Example: R -q -e "date()"
Create Tables In LaTeX, HTML, Markdown And ReStructuredText


-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
* https://rmarkdown.rstudio.com/lesson-7.html
                                              # It won't show the startup message when it is used in the command line.
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
                                              # Example: Rscript -e "date()"
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
</syntaxhighlight>
* 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


We can install littler using two ways.
== Create Word report ==
* 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 [http://dirk.eddelbuettel.com/code/littler.examples.html web version of examples] does not have the TOC.
=== Using the power of Word ===
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]


'''r''' was not meant to run interactively like '''R'''. See ''man r''.
=== knitr + pandoc ===
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* http://rmarkdown.rstudio.com/articles_docx.html


=== RInside: Embed R in C++ ===
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.
See [[R#RInside|RInside]]
<pre>
 
# Idea:
(''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.
#       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.


The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
Another way is
<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>


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'.
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


To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
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>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)
<pre>
<pre>
g++ -I/home/brb/Downloads/R-3.0.2/include \
knit("example.Rmd")
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
pandoc("example.md", format="epub")
    -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++.
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<pre>
<pre>
#include <RInside.h>                    // for the embedded R via RInside
> 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>
 
=== pander ===
Try pandoc[1] with a minimal reproducible example, you might give a try to my "[http://cran.r-project.org/web/packages/pander/ pander]" package [2] too:


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


    RInside R(argc, argv);              // create an embedded R instance
# http://johnmacfarlane.net/pandoc/
# http://rapporter.github.com/pander/
# http://rapporter.github.com/pander/#examples


    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
=== 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()


    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
This will download and install the current version of statconnDCOM


    exit(0);
You will need a working Internet connection
}
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
  argument is of length zero
</pre>
</pre>


The above can be compared to the Hello world example in Qt.
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
<pre>
 
#include <QApplication.h>
=== Convert from pdf to word ===
#include <QPushButton.h>
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 [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


int main( int argc, char **argv )
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
{
Package xtable will produce html output.
    QApplication app( argc, argv );
{{Pre}}
print(xtable(X), type="html")
</pre>


    QPushButton hello( "Hello world!", 0 );
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.
    hello.resize( 100, 30 );


    app.setMainWidget( &hello );
=== officer ===
    hello.show();
<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, "")


     return app.exec();
# 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>


=== [http://www.rfortran.org/ RFortran] ===
== Powerpoint ==
RFortran is an open source project with the following aim:
<ul>
<li>[https://cran.r-project.org/web/packages/officer/index.html officer] package  (formerly ReporteRs). [http://theautomatic.net/2020/07/28/how-to-create-powerpoint-reports-with-r/ How to create powerpoint reports with R]
</li>
<li>[https://davidgohel.github.io/flextable/ flextable] (imports '''officer''')
</li>
<li>[https://stackoverflow.com/a/21558466 R data.frame to table image for presentation].
<pre>
library(gridExtra)
grid.newpage()
grid.table(mydf)
</pre>
</li>
<li>[https://bookdown.org/yihui/rmarkdown/powerpoint-presentation.html Rmarkdown]
</li>
</ul>


''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
== PDF manipulation ==
[https://github.com/pridiltal/staplr staplr]


It works only on Windows platform with Microsoft Visual Studio installed:(
== R Graphs Gallery ==
* [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]


== Call R from other languages ==
== COM client or server ==
=== C ===
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


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]
=== Client ===


Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.


=== JRI ===
=== Server ===
http://www.rforge.net/JRI/
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


=== ryp2 ===
== Use R under proxy ==
http://rpy.sourceforge.net/rpy2.html
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


== Create a standalone Rmath library ==
== RStudio ==
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].
* [https://github.com/rstudio/rstudio Github]
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]


Here is my experience based on R 3.0.2 on Windows OS.
=== rstudio.cloud ===
https://rstudio.cloud/


=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
=== Launch RStudio ===
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.
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


=== Use Rmath library in our code ===
=== Create .Rproj file ===
<pre>
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.
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
With an RStudio project file, you can
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
* Restore .RData into workspace at startup
# It is OK to save the cpp file under any directory.
* 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


# Force to link against the static library <libRmath.a>
=== package search ===
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
https://github.com/RhoInc/CRANsearcher
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


# Force to link against dynamic library <Rmath.dll>
=== Git ===
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
</pre>
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]
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!
 
== 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]
 
== List files using regular expression ==
* Extension
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
<pre>
c:\R>RmathEx1
list.files(pattern = "^Something")
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>
</pre>


Below is the cpp program <RmathEx1.cpp>.
Using '''Sys.glob()"' as
<pre>
<pre>
//RmathEx1.cpp
> Sys.glob("~/Downloads/*.txt")
#define MATHLIB_STANDALONE
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
#include <iostream>
</pre>
#include "Rmath.h"


using std::cout; using std::cin; using std::endl;
== Hidden tool: rsync in Rtools ==
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe


int main()
sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
{
total size is 1198416  speedup is 3.71
  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 << ") = " <<
c:\Rtools\bin>
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
</pre>
</pre>


== Calling R.dll directly ==
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].
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.


== Create HTML report ==
== Install rgdal package (geospatial Data) on ubuntu ==
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor.
Terminal
{{Pre}}
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
</pre>


=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
R
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}}
install.packages("rgdal")
</pre>


* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
== Install sf package ==
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
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


=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
sudo apt update
* https://github.com/renkun-ken/formattable
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]


=== [https://github.com/crubba/htmltab htmltab] package ===
sudo apt install libgdal-dev # works on ubuntu 20.04 too
This package is NOT used to CREATE html report but EXTRACT html table.
                            # no need the previous lines
</pre>


=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
== Database ==
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.
* https://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 academic report ==
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
[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.
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


== Create pdf and epub files ==
'''Creating a new database''':
<syntaxhighlight lang='rsplus'>
{{Pre}}
# Idea:
library(DBI)
#        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.
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
dbDisconnect(mydb)
unlink("my-db.sqlite")


Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
# temporary database
<syntaxhighlight lang='bash'>
mydb <- dbConnect(RSQLite::SQLite(), "")
# Idea:
dbDisconnect(mydb)
#        knitr        pandoc
</pre>
#  rmd -------> md ----------> pdf


git clone https://github.com/yihui/knitr-examples.git
'''Loading data''':
cd knitr-examples
{{Pre}}
R -e "library(knitr); knit('001-minimal.Rmd')"
mydb <- dbConnect(RSQLite::SQLite(), "")
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
dbWriteTable(mydb, "mtcars", mtcars)
</syntaxhighlight>
dbWriteTable(mydb, "iris", iris)


To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
dbListTables(mydb)
<syntaxhighlight lang='rsplus'>
# Idea:
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


library(knitr)
dbListFields(con, "mtcars")
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</syntaxhighlight>
<pre>
## Windows OS, epub cannot be built
pandoc:
Error:
"source" (line 41, column 7):
unexpected "k"
expecting "{document}"


## Linux OS, epub missing figures and R codes.
dbReadTable(con, "mtcars")
## 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
'''Queries''':
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
Create Tables In LaTeX, HTML, Markdown And ReStructuredText


* https://rmarkdown.rstudio.com/lesson-7.html
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
* https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/
* [https://cran.r-project.org/web/packages/kableExtra/vignettes/awesome_table_in_html.html kableExtra] package


== Create Word report ==
res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
dbFetch(res)
</pre>


=== knitr + pandoc ===
'''Batched queries''':
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
{{Pre}}
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
dbClearResult(rs)
* http://rmarkdown.rstudio.com/articles_docx.html
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


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.
dbClearResult(rs)
<pre>
# Idea:
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</pre>
</pre>
and then
 
<pre>
'''Multiple parameterised queries''':
FILE <- "example"
{{Pre}}
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
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>
</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
'''Statements''':
<pre>
{{Pre}}
library(pander)
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
name = "demo"
#> [1] 0
knit(paste0(name, ".Rmd"), encoding = "utf-8")
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</pre>
</pre>


Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
* A pdf file: pandoc -s report.md -t latex -o report.pdf
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]
* 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
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
* 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!
=== [[MySQL#Use_through_R|RMySQL]] ===
<pre>
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
knit("example.Rmd")
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.
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)
=== MongoDB ===
<pre>
* http://www.r-bloggers.com/r-and-mongodb/
> pandoc("Rmd_to_Epub.md", format="epub")
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/
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)
=== odbc ===
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>


=== pander ===
=== RODBC ===
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>
=== DBI ===
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.


# http://johnmacfarlane.net/pandoc/
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
# http://rapporter.github.com/pander/
* To use databases with dplyr, you need to first install dbplyr
# http://rapporter.github.com/pander/#examples
* https://db.rstudio.com/dplyr/
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html


=== R2wd ===
'''Create a new SQLite database''':
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 <- read.csv("data/surveys.csv")
> library(R2wd)
plots <- read.csv("data/plots.csv")
> 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
my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)


You will need a working Internet connection
copy_to(my_db, surveys)
because installation needs to download a file.
copy_to(my_db, plots)
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
my_db
  argument is of length zero
</pre>
</pre>


The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
'''Connect to a database''':
{{Pre}}
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


=== Convert from pdf to word ===
library(dbplyr)
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")
</pre>


=== rtf ===
'''Querying the database with the SQL syntax''':
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
{{Pre}}
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
</pre>


=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
'''Querying the database with the dplyr syntax''':
Package xtable will produce html output. <syntaxhighlight lang='rsplus'>print(xtable(X), type="html")</syntaxhighlight>
{{Pre}}
surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)


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


=== [http://cran.r-project.org/web/packages/ReporteRs/index.html ReporteRs] ===
'''Simple database queries''':
Microsoft Word, Microsoft Powerpoint and HTML documents generation from R. The source code is hosted on https://github.com/davidgohel/ReporteRs
{{Pre}}
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</pre>


[https://statbandit.wordpress.com/2016/10/28/a-quick-exploration-of-reporters/ A quick exploration]
'''Laziness''' (instruct R to stop being lazy):
{{Pre}}
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</pre>


== PDF manipulation ==
'''Complex database queries''':
[https://github.com/pridiltal/staplr staplr]
{{Pre}}
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table


== R Graphs Gallery ==
surveys # The plot_id column also features in the surveys table
* [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].
# Join databases method 1
* 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.
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()
</pre>


== COM client or server ==
=== NoSQL ===
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]


=== Client ===
== Github ==


[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
=== R source  ===
https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.


=== Server ===
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


== Use R under proxy ==
=== R packages (only) source (metacran) ===
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.


== RStudio ==
=== Bioconductor packages source ===
* [https://github.com/rstudio/rstudio Github]
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]


=== rstudio.cloud ===
=== Send local repository to Github in R by using reports package ===
https://rstudio.cloud/
http://www.youtube.com/watch?v=WdOI_-aZV0Y


=== Launch RStudio ===
=== My collection ===
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
* 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


=== Create .Rproj file ===
=== How to download ===
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
Clone ~ Download.  
* Restore .RData into workspace at startup
* Command line
* Save workspace to .RData on exit
* Always save history (even if no saving .RData)
* etc
 
=== package search ===
https://github.com/RhoInc/CRANsearcher
 
=== Git ===
* (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]
 
== 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]
 
== List files using regular expression ==
* Extension
<pre>
<pre>
list.files(pattern = "\\.txt$")
git clone https://gist.github.com/4484270.git
</pre>
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
This will create a subdirectory called '4484270' with all cloned files there.
* Start with
 
* Within R
<pre>
<pre>
list.files(pattern = "^Something")
library(devtools)
source_gist("4484270")
</pre>
</pre>
 
or
Using '''Sys.glob()"' as
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
<pre>
> Sys.glob("~/Downloads/*.txt")
library(RJSONIO)
[1] "/home/brb/Downloads/ip.txt"       "/home/brb/Downloads/valgrind.txt"
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>
</pre>


== Hidden tool: rsync in Rtools ==
=== Jekyll ===
<pre>
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
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
== Connect R with Arduino ==
total size is 1198416  speedup is 3.71
* 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


c:\Rtools\bin>
== Android App ==
</pre>
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
And rsync works best when we need to sync folder.
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
<pre>
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]
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
== Common plots tips ==
total size is 8036311  speedup is 1.95
=== Create an empty plot ===
'''plot.new()'''   


c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
=== Overlay plots ===
cygwin warning:
[https://finnstats.com/index.php/2021/08/15/how-to-overlay-plots-in-r/ How to Overlay Plots in R-Quick Guide with Example].  
  MS-DOS style path detected: c:\users\limingc\Documents\binary\procexp.exe
<pre>
  Preferred POSIX equivalent is: /cygdrive/c/users/limingc/Documents/binary/procexp.exe
#Step1:-create scatterplot
  CYGWIN environment variable option "nodosfilewarning" turns off this warning.
plot(x1, y1)
  Consult the user's guide for more details about POSIX paths:
#Step 2:-overlay line plot
    http://cygwin.com/cygwin-ug-net/using.html#using-pathnames
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)
</pre>


c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
=== Save the par() and restore it ===
sending incremental file list
'''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'''.
binary/
* 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.
binary/procexp.exe
* If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
 
<pre>
sent 1767277 bytes received 35 bytes 3534624.00 bytes/sec
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
total size is 8036311  speedup is 4.55
old.par <- par(mar = c(5, 4, 4, 2) - 2)
 
## do plotting stuff with new settings
c:\Rtools\bin>
par(old.par)
</pre>
'''Example 2''': Use it inside a function with the [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/on.exit on.exit(0] function.
<pre>
ex <- function() {
  old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
  on.exit(par(old.par))
  ## ... do lots of par() settings and plots
  ## ...
  invisible() #-- now, par(old.par) will be executed
}
</pre>
'''Example 3''': It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.
<pre>
ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
</pre>
<pre>
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar
</pre>
</pre>


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
=== Grouped boxplots ===
* [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)


== Install rgdal package (geospatial Data) on ubuntu ==
=== [https://www.samruston.co.uk/ Weather Time Line] ===
Terminal
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].
<syntaxhighlight lang='bash'>
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
</syntaxhighlight>


R
=== Horizontal bar plot ===
<syntaxhighlight lang='rsplus'>
{{Pre}}
install.packages("rgdal")
library(ggplot2)
</syntaxhighlight>
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
 
                        "DDR", "BUM", "MAT", "HED", "EXP"),
== Install sf package ==
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
I got the following error even I have installed some libraries.  
ggplot(dtf, aes(x, y)) +
<pre>
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
checking GDAL version >= 2.0.1... no
  coord_flip() + xlab("") + ylab("Fold Change") 
configure: error: sf is not compatible with GDAL versions below 2.0.1
</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
[[:File:Ggplot2bar.svg]]
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
sudo apt install libgdal-dev
</syntaxhighlight>


== Set up Emacs on Windows ==
=== Include bar values in a barplot ===
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
<pre>
* [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.
(setq-default inferior-R-program-name
* [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]
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>


== Database ==
Use text().  
* 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]


=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
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].
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


'''Creating a new database''':
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.
<syntaxhighlight lang='rsplus'>
library(DBI)


mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
=== Grouped barplots ===
dbDisconnect(mydb)
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
unlink("my-db.sqlite")
{{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>
* [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)


# temporary database
=== Math expression ===
mydb <- dbConnect(RSQLite::SQLite(), "")
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
dbDisconnect(mydb)
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
</syntaxhighlight>
* 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")


'''Loading data''':
# Superscript
<syntaxhighlight lang='rsplus'>
plot(1:10, main = expression("My Title"^2))
mydb <- dbConnect(RSQLite::SQLite(), "")
# Subscript
dbWriteTable(mydb, "mtcars", mtcars)
plot(1:10, main = expression("My Title"[2]))
dbWriteTable(mydb, "iris", iris)


dbListTables(mydb)
# 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")


dbListFields(con, "mtcars")
# 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")


dbReadTable(con, "mtcars")
# Substituting Expressions
</syntaxhighlight>
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")
</pre>


'''Queries''':
=== Impose a line to a scatter plot ===
<syntaxhighlight lang='rsplus'>
* abline + lsfit # least squares
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')
{{Pre}}
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
</pre>
* abline + line # robust line fitting
{{Pre}}
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
</pre>
* lines
{{Pre}}
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')
</pre>


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
=== How to actually make a quality scatterplot in R: axis(), mtext() ===
[https://www.r-bloggers.com/2021/08/how-to-actually-make-a-quality-scatterplot-in-r/ How to actually make a quality scatterplot in R]


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
=== 3D scatterplot ===
* [http://sthda.com/english/wiki/scatterplot3d-3d-graphics-r-software-and-data-visualization Scatterplot3d: 3D graphics - R software and data visualization]. [https://stackoverflow.com/a/24510286 how to add legend to scatterplot3d in R] and consider '''xpd=TRUE'''.
* [[R_web#plotly|R web > plotly]]


res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
=== Rotating x axis labels for barplot ===
dbFetch(res)
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
</syntaxhighlight>
{{Pre}}
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</pre>
 
=== 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
{{Pre}}
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</pre>


'''Batched queries''':
=== Different colors of axis labels in barplot ===
<syntaxhighlight lang='rsplus'>
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]
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


dbClearResult(rs)
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
</syntaxhighlight>
{{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>


'''Multiple parameterised queries''':
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
<syntaxhighlight lang='rsplus'>
{{Pre}}
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
barplot(tN, col = rainbow(20), axisnames = F)
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
nrow(dbFetch(rs))
</pre>
#> [1] 4
dbClearResult(rs)
</syntaxhighlight>


'''Statements''':
=== Use text() to draw labels on X/Y-axis including rotation ===
<syntaxhighlight lang='rsplus'>
* 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]
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
* [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).
#> [1] 0
** par("usr") is determined *after* a plot has been created
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
dbBind(rs, param = list(x = 4.5))
* https://datascienceplus.com/building-barplots-with-error-bars/
dbGetRowsAffected(rs)
{{Pre}}
#> [1] 4
par(mar = c(5, 6, 4, 5) + 0.1)
dbClearResult(rs)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
</syntaxhighlight>
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</pre>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/
 
=== Vertically stacked plots with the same x axis ===
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in


=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
=== Include labels on the top axis/margin: axis() and mtext() ===
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>
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]]


=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
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>


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


=== MongoDB ===
=== Legend tips ===
* http://www.r-bloggers.com/r-and-mongodb/
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


=== odbc ===
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
{{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>


=== RODBC ===
'''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].
<pre>
legend("bottomright", inset=.05, )
</pre>


=== DBI ===
'''legend without a box'''
<pre>
legend(, bty = "n")
</pre>


=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
'''Add a legend title'''
* To use databases with dplyr, you need to first install dbplyr
<pre>
* https://db.rstudio.com/dplyr/
legend(, title = "")
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
</pre>
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html


'''Create a new SQLite database''':
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.
<syntaxhighlight lang='rsplus'>
 
surveys <- read.csv("data/surveys.csv")
=== Superimpose a density plot or any curves ===
plots <- read.csv("data/plots.csv")
Use '''lines()'''.  


my_db_file <- "portal-database.sqlite"
Example 1
my_db <- src_sqlite(my_db_file, create = TRUE)
{{Pre}}
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))


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


'''Connect to a database''':
Example 2
<syntaxhighlight lang='rsplus'>
{{Pre}}
download.file(url = "https://ndownloader.figshare.com/files/2292171",
require(survival)
              destfile = "portal_mammals.sqlite", mode = "wb")
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
</pre>


library(dbplyr)
Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")
</syntaxhighlight>


'''Querying the database with the SQL syntax''':
=== log scale ===
<syntaxhighlight lang='rsplus'>
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).
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
</syntaxhighlight>


'''Querying the database with the dplyr syntax''':
[[:File:Logscale.png]]
<syntaxhighlight lang='rsplus'>
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
=== Custom scales ===
</syntaxhighlight>
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]


'''Simple database queries''':
== Time series ==
<syntaxhighlight lang='rsplus'>
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
surveys %>%
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</syntaxhighlight>


'''Laziness''' (instruct R to stop being lazy):
=== 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])
data_subset <- 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]
  filter(weight < 5) %>%
* https://timelyportfolio.github.io/rCharts_time_series/history.html
  select(species_id, sex, weight) %>%
  collect()
</syntaxhighlight>


'''Complex database queries''':
{{Pre}}
<syntaxhighlight lang='rsplus'>
library(quantmod)
plots <- tbl(mammals, "plots")
getSymbols("AAPL")
plots # # The plot_id column features in the plots table
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')


surveys # The plot_id column also features in the surveys table
tail(Cl(DJI))
</pre>


# Join databases method 1
=== tidyquant: Getting stock data ===
plots %>%
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()
</syntaxhighlight>


=== NoSQL ===
=== Timeline plot ===
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R


== Github ==
=== Clockify ===
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]


=== R source  ===
== Circular plot ==
https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
* 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.


If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
== Word cloud ==
* [http://www.sthda.com/english/wiki/text-mining-and-word-cloud-fundamentals-in-r-5-simple-steps-you-should-know Text mining and word cloud fundamentals in R : 5 simple steps you should know]
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
* [https://github.com/lepennec/ggwordcloud?s=09 ggwordcloud]


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


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


=== Send local repository to Github in R by using reports package ===
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
http://www.youtube.com/watch?v=WdOI_-aZV0Y
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]


=== My collection ===
=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
* https://github.com/arraytools
* [https://blog.rstudio.com/2015/05/01/rstudio-v0-99-preview-graphviz-and-diagrammer/ Graphviz and DiagrammeR]
* https://gist.github.com/4383351 heatmap using leukemia data
* http://rich-iannone.github.io/DiagrammeR/,
* https://gist.github.com/4382774 heatmap using sequential data
** [http://rich-iannone.github.io/DiagrammeR/io.html#r-markdown rmarkdown]
* https://gist.github.com/4484270 biocLite
** [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]


=== How to download ===
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams


Clone ~ Download.  
=== DAGitty (browser-based and R package) ===
* Command line
* http://dagitty.net/
<pre>
* https://cran.r-project.org/web/packages/dagitty/index.html
git clone https://gist.github.com/4484270.git
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


* Within R
=== dagR ===
<pre>
* https://cran.r-project.org/web/packages/dagR
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 ===
=== Gmisc ===
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]


== Connect R with Arduino ==
=== Concept Maps ===
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.
* 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 ==
=== flow ===
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
[https://cran.r-project.org/web/packages/flow/ flow], [https://predictivehacks.com/?all-tips=how-to-draw-flow-diagrams-in-r How To Draw Flow Diagrams In R]
* [http://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 ==
== Venn Diagram ==
=== Grouped boxplots ===
[[Venn_diagram|Venn diagram]]
* [http://sphaerula.com/legacy/R/boxplotTwoWay.html Box Plots of Two-Way Layout]
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
** 'at' parameter in boxplot() to change the equal spaced boxplots
** embed par(mar=) in boxplot()
** mtext(line=) to solve the problem the xlab overlapped with labels.
* [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others ggplot2 approach] (Hint: '''facet_grid''' is used)


=== [https://www.samruston.co.uk/ Weather Time Line] ===
== hexbin plot ==
The plot looks similar to a boxplot though it is not. See a [https://www.samruston.co.uk/images/screens/screen_2.png screenshot] on Android by [https://www.samruston.co.uk/ Sam Ruston].
* [https://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.


=== Horizontal bar plot ===
== Bump chart/Metro map ==
<syntaxhighlight lang='rsplus'>
https://dominikkoch.github.io/Bump-Chart/
library(ggplot2)
 
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
== Amazing/special plots ==
                        "DDR", "BUM", "MAT", "HED", "EXP"),
See [[Amazing_plot|Amazing plot]].
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
 
ggplot(dtf, aes(x, y)) +
== Google Analytics ==
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
=== GAR package ===
  coord_flip() + xlab("") + ylab("Fold Change") 
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html
</syntaxhighlight>
 
== Linear Programming ==
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/


[[:File:Ggplot2bar.svg]]
== 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.


=== Include bar values in a barplot ===
== Amazon Alexa ==
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
* [http://stackoverflow.com/questions/12481430/how-to-display-the-frequency-at-the-top-of-each-factor-in-a-barplot-in-r barplot(), text() and axis()] functions. The data can be from a table() object.
* [https://stackoverflow.com/questions/11938293/how-to-label-a-barplot-bar-with-positive-and-negative-bars-with-ggplot2 How to label a barplot bar with positive and negative bars with ggplot2]


Use text().  
== R and Singularity ==
https://rviews.rstudio.com/2017/03/29/r-and-singularity/


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].
== Teach kids about R with Minecraft ==
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html


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


=== Grouped barplots ===
== Credentials and secrets ==
* 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://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]
library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
      geom_bar(position="dodge", stat="identity")
</syntaxhighlight>
* https://datascienceplus.com/building-barplots-with-error-bars/. The error bars define 2 se (95% interval) for the black-and-white version and 1 se (68% interval) for ggplots. Be careful.<syntaxhighlight lang='rsplus'>
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</syntaxhighlight>
* [http://stackoverflow.com/questions/27466035/adding-values-to-barplot-of-table-in-r two bars in one factor] (stack). The data can be a 2-dim matrix with numerical values.
* [http://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r two bars in one factor], [https://stats.stackexchange.com/questions/14118/drawing-multiple-barplots-on-a-graph-in-r Drawing multiple barplots on a graph in R] (next to each other)
** [https://datascienceplus.com/building-barplots-with-error-bars/ Include error bars]
* [http://bl.ocks.org/patilv/raw/7360425/ Three variables] barplots
* [https://peltiertech.com/stacked-bar-chart-alternatives/ More alternatives] (not done by R)


=== Math expression ===
== Hide a password ==
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
=== keyring package ===
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* https://cran.r-project.org/web/packages/keyring/index.html
* https://andyphilips.github.io/blog/2017/08/16/mathematical-symbols-in-r-plots.html
* [http://theautomatic.net/2019/06/25/how-to-hide-a-password-in-r-with-the-keyring-package/ How to hide a password in R with the Keyring package]
** 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
=== getPass ===
# '~' is to add space and '*' is to squish characters together
[https://cran.r-project.org/web/packages/getPass/README.html getPass]
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
== Vision and image recognition ==
plot(x,y,
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
* [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
    ylab = expression(paste("Here is some text of ", phi^{rho})),
    main = "Expressions with Text")


# Substituting Expressions
== Creating a Dataset from an Image ==
plot(x,y,
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]
    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 ===
== Turn pictures into coloring pages ==
* abline + lsfit # least squares
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
: <syntaxhighlight lang='rsplus'>
 
plot(cars)
== Numerical optimization ==
abline(lsfit(cars[, 1], cars[, 2]))
[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]
# 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 ===
* [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://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
<syntaxhighlight lang='rsplus'>
* [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.
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
</syntaxhighlight>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.


=== Different colors of axis labels in barplot ===
== Ryacas: R Interface to the 'Yacas' Computer Algebra System ==
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]
[https://blog.ephorie.de/doing-maths-symbolically-r-as-a-computer-algebra-system-cas Doing Maths Symbolically: R as a Computer Algebra System (CAS)]


Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
== Game ==
<syntaxhighlight lang='rsplus'>
* [https://kbroman.org/miner_book/?s=09 R Programming with Minecraft]
tN <- table(Ni <- stats::rpois(100, lambda = 5))
* [https://cran.r-project.org/web/packages/pixelpuzzle/index.html pixelpuzzle]
r <- barplot(tN, col = rainbow(20))
* [https://www.rostrum.blog/2022/09/24/pixeltrix/ Interactive pixel art in R with {pixeltrix}]
axis(1, 1, LETTERS[1], col.axis="red", col="red")
* [https://rtaoist.blogspot.com/2021/03/r-shiny-maths-games-for-6-years-old.html Shiny math game]
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
* [https://cran.microsoft.com/web/packages/mazing/index.html mazing]: Utilities for Making and Plotting Mazes
</syntaxhighlight>
* [https://github.com/jeroenjanssens/raylibr/blob/main/demo/snake.R snake] which is based on [https://github.com/jeroenjanssens/raylibr raylibr]


Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
== Music ==
<syntaxhighlight lang='rsplus'>
* [https://flujoo.github.io/gm/ gm]. Require to install [https://musescore.org/en MuseScore], an open source and free notation software.
barplot(tN, col = rainbow(20), axisnames = F)
 
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
== SAS ==
</syntaxhighlight>
[https://github.com/MangoTheCat/sasMap sasMap] Static code analysis for SAS scripts


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


=== Vertically stacked plots with the same x axis ===
= Tricks =
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in


=== Increase/decrease legend font size ===
== Getting help ==
https://stackoverflow.com/a/36842578
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
<syntaxhighlight lang='rsplus'>
* https://stat.ethz.ch/pipermail/r-help/
plot(rnorm(100))
* https://stat.ethz.ch/pipermail/r-devel/
op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1)
par(op)
</syntaxhighlight>


=== Superimpose a density plot or any curves ===
== Better Coder/coding, best practices ==
Use '''lines()'''.  
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]
* [https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.9169?campaign=woletoc Best practices in statistical computing] Sanchez 2021


Example 1
== [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
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))
</syntaxhighlight>


Example 2
== Getting user's home directory ==
<syntaxhighlight lang='rsplus'>
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?]
require(survival)
{{Pre}}
n = 10000
# Windows
beta1 = 2; beta2 = -1
normalizePath("~")   # "C:\\Users\\brb\\Documents"
lambdaT = 1 # baseline hazard
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
lambdaC = 2  # hazard of censoring
Sys.getenv("HOME"# "C:/Users/brb/Documents"
set.seed(1234)
 
x1 = rnorm(n,0)
# Mac
x2 = rnorm(n,0)
normalizePath("~")   # [1] "/Users/brb"
# true event time
Sys.getenv("R_USER") # [1] ""
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))
Sys.getenv("HOME")   # "/Users/brb"
C <- rweibull(n, shape=1, scale=lambdaC)   
time = pmin(T,C
status <- 1*(T <= C)
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1),
    ylab="Survival probability",
    main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
</syntaxhighlight>


=== Custom scales ===
# Linux
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
normalizePath("~")  # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # [1] "/home/brb"
</pre>


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


=== Time series stock price plot ===
== Distinguish Windows and Linux/Mac, R.Version() ==
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.
* [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'>
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
library(quantmod)
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
getSymbols("AAPL")
<pre>
getSymbols("IBM") # similar to AAPL
get_os <- function(){
getSymbols("CSCO") # much smaller than AAPL, IBM
  sysinf <- Sys.info()
getSymbols("DJI") # Dow Jones, huge
  if (!is.null(sysinf)){
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
    os <- sysinf['sysname']
    col='orange', subset = '2017::2017-08')
    if (os == 'Darwin')
      os <- "osx"
  } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
</pre>
<pre>
names(R.Version())
#  [1] "platform"      "arch"          "os"            "system"       
#  [5] "status"        "major"          "minor"          "year"         
#  [9] "month"          "day"            "svn rev"        "language"     
# [13] "version.string" "nickname"  
getRversion()
# [1] ‘4.3.0’
</pre>


tail(Cl(DJI))
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
</syntaxhighlight>
* 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]


=== Timeline plot ===
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://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
<pre>
R_LIBS_SITE=F:/R/library
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''. In R, run '''Sys.getenv("R_LIBS_SITE")''' or '''Sys.getenv("R_LIBS_USER")''' to query the environment variable. See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/EnvVar.html Environment Variables].


== Circular plot ==
=== What is the best place to save Rconsole on Windows platform ===
* http://freakonometrics.hypotheses.org/20667 which uses https://cran.r-project.org/web/packages/circlize/ circlize] package.
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.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 ==
My preferred settings:
* [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]
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
* Size: 12
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
* background: black
* normaltext: white
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow


== World map ==
and others (default options)
[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)
* pagebg: white
* pagetext: navy
* highlight: DarkRed
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].


=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
=== How R starts up ===
* http://rich-iannone.github.io/DiagrammeR/
https://rstats.wtf/r-startup.html
* https://donlelek.github.io/2015-03-31-dags-with-r/


=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
=== startup - Friendly R Startup Configuration ===
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
https://github.com/henrikbengtsson/startup


=== DAGitty (browser-based and R package) ===
== Saving and loading history automatically: .Rprofile & local() ==
* http://dagitty.net/
<ul>
* https://cran.r-project.org/web/packages/dagitty/index.html
<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)


=== dagR ===
local({r <- getOption("repos")
* https://cran.r-project.org/web/packages/dagR
      r["CRAN"] <- "https://cran.rstudio.com"
      options(repos=r)})


== Venn Diagram ==
.First <- function(){
* limma http://www.ats.ucla.edu/stat/r/faq/venn.htm - only black and white?
# library(tidyverse)
* VennDiagram - input has to be the numbers instead of the original vector?
cat("\nWelcome at", date(), "\n")
* http://manuals.bioinformatics.ucr.edu/home/R_BioCondManual#TOC-Venn-Diagrams and the [http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R R code] or the [http://www.bioconductor.org/packages/release/bioc/html/systemPipeR.html Bioc package systemPipeR]
}
<syntaxhighlight lang='rsplus'>
# systemPipeR package method
library(systemPipeR)
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
OLlist <- overLapper(setlist[1:3], type="vennsets")
vennPlot(list(OLlist))                           


# R script source method
.Last <- function(){
source("http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R")
cat("\nGoodbye at ", date(), "\n")
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))
</pre>
setlist <- structure(list(A = c("o", "h", "u", "p", "i", "s", "a", "w",
<li>https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
"b", "z", "n", "c", "k", "j", "y", "m", "t", "q"), B = c("h",
<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.
"r", "x", "y", "b", "t", "d", "o", "m", "q", "g", "v", "c", "u",
<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)
"f", "z"), C = c("b", "e", "t", "u", "s", "j", "o", "k", "d",
</ul>
"l", "g", "i", "w", "n", "p", "a", "y", "x", "m", "z"), D = c("f",
'''Linux''' or '''Mac'''
"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")
In '''~/.profile''' or '''~/.bashrc''' I put:
counts <- list(sapply(OLlist$Venn_List, length)) 
<pre>
vennPlot(counts=counts)                         
export R_HISTFILE=~/.Rhistory
</syntaxhighlight>
</pre>
 
In '''~/.Rprofile''' I put:
[[:File:Vennplot.png]]
<pre>
 
if (interactive()) {
== Bump chart/Metro map ==
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))
https://dominikkoch.github.io/Bump-Chart/
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>


== Amazing plots ==
'''Windows'''
See [[Amazing_plot|Amazing plot]].


== Google Analytics ==
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.
=== GAR package ===
<pre>
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html
if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>


== Linear Programming ==
== Disable "Save workspace image?" prompt when exit R? ==
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in R?]


== Linear Algebra ==
== R release versions ==
* [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.
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
* [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 ==
== getRversion() ==
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
<pre>
getRversion()
[1] ‘4.3.0’
</pre>


== R and Singularity ==
== Detect number of running R instances in Windows ==
https://www.rstudio.com/rviews/2017/03/29/r-and-singularity/
* 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.


== Teach kids about R with Minecraft ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html


== Secure API keys ==
Image Name                    PID Session Name        Session#    Mem Usage
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
============================================================================
Rgui.exe                      1096 Console                    1    44,712 K


== Hide a password: keyring package ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
* https://cran.r-project.org/web/packages/keyring/index.html
* [http://theautomatic.net/2019/06/25/how-to-hide-a-password-in-r-with-the-keyring-package/ How to hide a password in R with the Keyring package]


== Vision and image recognition ==
Image Name                    PID Session Name        Session#    Mem Usage
* 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
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"
 
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
</pre>


== Turn pictures into coloring pages ==
== Editor ==
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs


== Numerical optimization ==
<ul>
* [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]
<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).
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
* [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.
<pre>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
(setq-default inferior-R-program-name
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
</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


= R packages =
== GUI for Data Analysis ==
[[R_packages|R packages]]
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023


= Tricks =
=== BlueSky Statistics ===
* https://www.blueskystatistics.com/Default.asp
* [https://r4stats.com/articles/software-reviews/bluesky/ A Comparative Review of the BlueSky Statistics GUI for R]


== Getting help ==
=== Rcmdr ===
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
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.
* https://stat.ethz.ch/pipermail/r-help/
* https://stat.ethz.ch/pipermail/r-devel/


== Better Coder/coding, best practices ==
=== Deducer ===
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
http://cran.r-project.org/web/packages/Deducer/index.html
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]


== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
=== jamovi ===
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
* https://www.jamovi.org/
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]


== Getting user's home directory ==
== Scope ==
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?]
See  
<syntaxhighlight lang='rsplus'>
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
# Windows
normalizePath("~")  # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")  # "C:/Users/brb/Documents"


# Mac
=== source() ===
normalizePath("~")   # [1] "/Users/brb"
* [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.
Sys.getenv("R_USER") # [1] ""
* [[#How_to_exit_a_sourced_R_script|source()]] does not work like C's preprocessor where statements in header files will be literally inserted into the code. It does not work when you define a variable in a function but want to use it outside the function (even through '''source()''')
Sys.getenv("HOME")   # "/Users/brb"


# Linux
{{Pre}}
normalizePath("~")  # [1] "/home/brb"
## foo.R ##
Sys.getenv("R_USER") # [1] ""
cat(ArrayTools, "\n")
Sys.getenv("HOME")   # [1] "/home/brb"
## End of foo.R
</syntaxhighlight>


== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
# 1. Error
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
predict <- function() {
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
  ArrayTools <- "C:/Program Files" # or through load() function
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
  source("foo.R")                  # or through a function call; foo()
* [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]
predict()  # Object ArrayTools not found


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
# 2. OK. Make the variable global
<pre>
predict <- function() {
R_LIBS_SITE=F:/R/library
  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()
</pre>
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''.


Note that on Windows OS, R/etc contains
'''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>
$ ls -l /c/Progra~1/r/r-3.2.0/etc
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
total 142
> ttt
-rw-r--r--   1  Administ     1043 Jun 20  2013 Rcmd_environ
  type JpnTest
-rw-r--r--   1  Administ     1924 Mar 17  2010 Rconsole
1   a    999
-rw-r--r--    Administ     943 Oct  3 2011 Rdevga
2   b    999
-rw-r--r--   1  Administ      589 May 20  2013 Rprofile.site
3    c    999
-rw-r--r--   1  Administ   251894 Jan 17  2015 curl-ca-bundle.crt
4   d     999
drwxr-xr-x   1  Administ        0 Jun  8 10:30 i386
5   e     999
-rw-r--r--   1  Administ     1160 Dec 31  2014 repositories
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
-rw-r--r--   1  Administ   30188 Mar 17  2010 rgb.txt
> jpntest()
drwxr-xr-x   3  Administ        0 Jun  8 10:30 x64
   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>


$ ls /c/Progra~1/r/r-3.2.0/etc/i386
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.
Makeconf


$ cat /c/Progra~1/r/r-3.2.0/etc/Rconsole
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
# 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
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
# This can be `yes' (for MDI) or `no' (for SDI).
  MDI = yes
# MDI = no


# the next two are only relevant for MDI
=== New environment ===
toolbar = yes
* http://adv-r.had.co.nz/Environments.html.
statusbar = no
* [https://www.r-bloggers.com/2011/06/environments-in-r/ Environments in R]
* load(), attach(), with().
* [https://stackoverflow.com/questions/33109379/how-to-switch-to-a-new-environment-and-stick-into-it How to switch to a new environment and stick into it?] seems not possible!


## Font.
Run the same function on a bunch of R objects
# Please use only fixed width font.
{{Pre}}
# If font=FixedFont the system fixed font is used; in this case
mye = new.env()
# points and style are ignored. If font begins with "TT ", only
load(<filename>, mye)
# True Type fonts are searched for.
for(n in names(mye)) n = as_tibble(<nowiki>mye[[n]]</nowiki>)
font = TT Courier New
</pre>
points = 10
style = normal # Style can be normal, bold, italic


# Dimensions (in characters) of the console.
Just look at the contents of rda file without saving to anywhere (?load)
rows = 25
<pre>
columns = 80
local({
# Dimensions (in characters) of the internal pager.
  load("myfile.rda")
pgrows = 25
  ls()
pgcolumns = 80
})
# should options(width=) be set to the console width?
</pre>
setwidthonresize = yes
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}}
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")
</pre>
If we want to read data from internet, '''load()''' works but not attach().
<pre>
con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840
</pre>
[https://stackoverflow.com/a/39621091 source() case].
<pre>
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>


# memory limits for the console scrolling buffer, in chars and lines
=== str( , max) function ===
# NB: bufbytes is in bytes for R < 2.7.0, chars thereafter.
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]
bufbytes = 250000
buflines = 8000


# Initial position of the console (pixels, relative to the workspace for MDI)
If we use str() on a function like str(lm), it is equivalent to args(lm)
# xconsole = 0
# yconsole = 0


# Dimension of MDI frame in pixels
For a complicated list object, it is useful to use the '''max.level''' argument; e.g. str(, max.level = 1)
# 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 large data frame, we can use the '''tibble()''' function; e.g. mydf %>% tibble()
# or in multiple windows (one for each topic)
# pagerstyle can be set to `singlewindow' or `multiplewindows'
pagerstyle = multiplewindows


## Colours for console and pager(s)
=== tidy() function ===
# (see rwxxxx/etc/rgb.txt for the known colours).
broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See [[Tidyverse#broom|here]].
background = White
normaltext = NavyBlue
usertext = Red
highlight = DarkRed


## Initial position of the graphics window
=== View all objects present in a package, ls() ===
## (pixels, <0 values from opposite edge)
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
xgraphics = -25
{{Pre}}
ygraphics = 0
> 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"


## Language for messages
> ls("package:mypackage")
language =
[1] "_mypackage_rcpp_hello_world" "evalCpp"                    "library.dynam.unload"     
 
[4] "rcpp_hello_world"            "system.file"
## Default setting for console buffering: 'yes' or 'no'
buffered = yes
</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)
# ## Example .Renviron on Windows
# [1] TRUE
# R_LIBS=C:/R/library
c(T,F,T) && c(F,T,T)
# MY_TCLTK="c:/Program Files/Tcl/bin"
# [1] FALSE
 
c(T,F,T) && c(NA,T,T)
# ## Example of setting R_DEFAULT_PACKAGES (from R CMD check)
# [1] NA
# R_DEFAULT_PACKAGES='utils,grDevices,graphics,stats'
</pre>
# # this loads the packages in the order given, so they appear on
<pre>
# # the search path in reverse order.
# Assume 'b' is not defined
brb@brb-T3500:~$
> 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>
</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))


=== What is the best place to save Rconsole on Windows platform ===
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
</pre>
 
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)
== for-loop, control flow ==
* pagebg: white
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Control ?Control]
* pagetext: navy
* '''next''' can be used to skip the rest of the inner-most loop
* highlight: DarkRed
* [https://www.programiz.com/r/ifelse-function ifelse() Function]
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].
== Vectorization ==
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
* [https://github.com/henrikbengtsson/matrixstats matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors). E.g. col / rowMedians(), col / rowRanks(), and col / rowSds(). [https://github.com/HenrikBengtsson/matrixStats/wiki/Benchmark-reports Benchmark reports].


== Saving and loading history automatically: .Rprofile & local() ==
=== sapply vs vectorization ===
* http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
* '''.Rprofile''' will automatically be loaded when R has started from that directory
* Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
* '''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
* [http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]  
** You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
** At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
options(continue="  ") # default is "+ "
options(editor="nano") # default is "vi" on Linux
# options(htmlhelp=TRUE)


local((r <- getOption("repos")
=== lapply vs for loop ===
  r["CRAN"] <- "http://cran.rstudio.com"
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
  options(repos = r)))
* https://code-examples.net/en/q/286e03a
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]


.First <- function(){
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
# library(Hmisc)
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?]
  cat("\nWelcome at", date(), "\n")
* Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
}
{{Pre}}
split(x, f, drop = FALSE, …)
</pre>
* [https://stackoverflow.com/a/3321659 Split a vector into chunks]. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the '''split() + lapply() + do.call()''' or '''split() + sapply()''' operations.
<pre>
d <- 1:10
chunksize <- 4
ceiling(1:10/4)
# [1] 1 1 1 1 2 2 2 2 3 3
split(d, ceiling(seq_along(d)/chunksize))
# $`1`
# [1] 1 2 3 4
#
# $`2`
# [1] 5 6 7 8
#
# $`3`
# [1] 9 10
do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) )  
#  1  2  3
# 10 26 19


.Last <- function(){
# bigmemory vignette
cat("\nGoodbye at ", date(), "\n")
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
</pre>  
                    function(i) birthmonth(x[i, c('Year','Month'),
* https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
                                            drop=FALSE]))
* The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
</pre>
* [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
* Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.  
{{Pre}}
split(mtcars,mtcars$cyl)


'''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
</pre>


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.
=== strsplit and sapply ===
<pre>
{{Pre}}
if (interactive()) {
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
> strsplit(namedf, " ")
}
[[1]]
</pre>
[1] "John" "ABC"


== R release versions ==
[[2]]
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
[1] "Mary" "CDE"


== Detect number of running R instances in Windows ==
[[3]]
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
[1] "Kat" "FGH"
<pre>
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"
> sapply(strsplit(namedf, " "), "[", 1)
[1] "John" "Mary" "Kat"
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"
</pre>


Image Name                    PID Session Name        Session#    Mem Usage
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
============================================================================
<ul>
Rgui.exe                      1096 Console                    1     44,712 K
<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


C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
# 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)))
</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


Image Name                    PID Session Name        Session#    Mem Usage
apply(x, 1, mean, na.rm=T)
============================================================================
# [1] 31 27 28 29 30 31 32 33 34 35
Rserve.exe                    6108 Console                    1   381,796 K
</pre>
</pre>
In R, we can use
</li>
<pre>
<li>[https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
</li>
[1] ""                                                                          
<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]
[2] "Image Name                    PID Session Name        Session#   Mem Usage"
</li>
[3] "============================================================================"
</ul>
[4] "Rgui.exe                      1096 Console                    1     44,804 K"
 
=== 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


> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
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
</pre>
</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]]


== Editor ==
</li>
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
<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}}
* 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).
group <- rownames(x)
* [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]
rowsum(x, group, na.rm=T)/as.vector(table(group))
* [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).
#  [,1] [,2] [,3] [,4] [,5] [,6]
* [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].
# a 10.0 20.0 30.0 40.0 50.0 60.0
* 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
# 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
== GUI for Data Analysis ==
# d  7.5 17.5 27.5 37.5 47.5 57.5
 
</pre>
=== Rcmdr ===
</li>
http://cran.r-project.org/web/packages/Rcmdr/index.html
</ul>
 
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
=== Deducer ===
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
http://cran.r-project.org/web/packages/Deducer/index.html
* [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)


=== jamovi ===
# Another example: select rows with a minimum value from a certain column (yval in this case)
* https://www.jamovi.org/
> mydf <- read.table(header=T, text='
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
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
</pre>
</li>
</ul>


== Scope ==
=== Mean by Group ===
See
[https://statisticsglobe.com/mean-by-group-in-r Mean by Group in R (2 Examples) | dplyr Package vs. Base R]
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
<pre>
* [[#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()''')
aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                          # Specify function (i.e. mean)
</pre>
<pre>
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                        # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
              list(name = mean))              # Specify function
</pre>
* [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/ave ave(x, ..., FUN)],
* aggregate(x, by, FUN),
* by(x, INDICES, FUN): return is a list
* tapply(): return results as a matrix or array. Useful for [https://en.wikipedia.org/wiki/Jagged_array ragged array].


<syntaxhighlight lang='rsplus'>
== Apply family ==
## foo.R ##
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.  
cat(ArrayTools, "\n")
## End of foo.R


# 1. Error
The following list gives a hierarchical relationship among these functions.
predict <- function() {
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
  ArrayTools <- "C:/Program Files" # or through load() function  
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
  source("foo.R")                 # or through a function call; foo()
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
}
*** '''replicate'''(n, expr, simplify = "array")
predict()   # Object ArrayTools not found
** '''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


# 2. OK. Make the variable global
[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?]
predict <- function() {
* 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
  ArrayTools <<- "C:/Program Files'
* lapply - When you want to apply a function to each element of a list in turn and get a list back.
  source("foo.R")
* 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.
predict() 
ArrayTools


# 3. OK. Create a global variable
Some short examples:
ArrayTools <- "C:/Program Files"
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
predict <- function() {
* [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.
  source("foo.R")
* [https://stackoverflow.com/a/7141669 How to use which one (apply family) when?]
}
 
predict()
=== Apply vs for loop ===
</syntaxhighlight>
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`?]
 
=== Progress bar ===
[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?]
 
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://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]
 
=== simplify option in sapply() ===
<pre>
library(KEGGREST)
 
names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2


'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
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
</pre>


Example 1.  
=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
* 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>
<pre>
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
z <- mapply(function(u, v) { max(u, v) },
> ttt
            u = x[, 1], v = x[, 2])
  type JpnTest
</pre>
1    a    999
</li>
2    b    999
<li>[http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming </li>
3    c    999
<li>Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
4    d    999
<ul>
5    e    999
<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().
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
{{Pre}}
> jpntest()
# Syntax: Map(f, ...)
  type JpnTest
1    a      N5
2    b    999
3    c    999
4    d    999
5    e    999
> ttt
  type JpnTest
1    a    999
2    b    999
3    c    999
4    d    999
5    e    999
</pre>


Example 2. [http://stackoverflow.com/questions/1236620/global-variables-in-r How can we set global variables inside a function?] The answer is to use the "<<-" operator or '''assign(, , envir = .GlobalEnv)''' function.
xs <- replicate(5, runif(10), simplify = FALSE)
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


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


Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
> m1 <- data.frame(id=letters[1:4], val=1:4)
 
> m2 <- data.frame(id=letters[2:6], val=2:6)
=== New environment ===
> merge(m1, m2, "id", all = T)
http://adv-r.had.co.nz/Environments.html
  id val.x val.y
 
1  a    1    NA
Run the same function on a bunch of R objects
2  b    2    2
<syntaxhighlight lang='rsplus'>
3 c    3    3
mye = new.env()
4  d    4    4
load(<filename>, mye)
5  e    NA    5
for(n in names(mye)) n = as_tibble(mye[[n]])
6  f    NA    6
</syntaxhighlight>
> 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]


=== View all objects present in a package, ls() ===
=== sapply & vapply ===
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
<syntaxhighlight lang='rsplus'>
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
> devtools::load_all("mypackage")
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
> search()
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
[1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
* 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.
[4] "package:stats"    "package:graphics"  "package:grDevices"
[7] "package:utils"    "package:datasets"  "package:methods"
[10] "Autoloads"        "package:base"


> ls("package:mypackage")
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].
[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
=== rapply - recursive version of lapply ===
* an integer (the position in the ‘search’ list);
* http://4dpiecharts.com/tag/recursive/
* the character string name of an element in the search list;
* [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].
* an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).


== Speedup R code ==
=== replicate ===
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
{{Pre}}
> replicate(5, rnorm(3))
          [,1]      [,2]      [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</pre>


== Profiler ==
See [[#parallel_package|parSapply()]] for a parallel version of replicate().
(Video) [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler]


== && vs & ==
=== Vectorize ===
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  
* [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/Vectorize Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE)]: creates a function wrapper that vectorizes a scalar function. Its value is a list or vector or array. It calls '''mapply()'''.
{{Pre}}
> rep(1:4, 4:1)
[1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
[[1]]
[1] 1 1 1 1


The 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.
[[2]]
[1] 2 2 2


== stopifnot(): function argument sanity check ==
[[3]]
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]
[1] 3 3


== Vectorization ==
[[4]]
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
[1] 4
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
</pre>
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy]
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
{{Pre}}
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html  
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
[1] 2.17123
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
[1] 1.6491761 0.9610109
* [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].
</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


=== sapply vs vectorization ===
myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
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>


=== lapply vs for loop ===
== plyr and dplyr packages ==
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
* https://code-examples.net/en/q/286e03a
* [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() ===
[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
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?]
* Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
:<syntaxhighlight lang='rsplus'>
split(x, f, drop = FALSE, …)
</syntaxhighlight>
* Split rows of a data frame/matrix <syntaxhighlight lang='rsplus'>
split(mtcars,mtcars$cyl)
</syntaxhighlight>
* Split columns of a data frame/matrix. <syntaxhighlight lang='rsplus'>
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10)))
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
</syntaxhighlight>
* split() + sapply() to merge columns. See below [[#Mean_of_duplicated_columns:_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)))
[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.
# 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 ===
# plyr has a common syntax -- easier to remember
<syntaxhighlight lang='rsplus'>
# plyr requires less code since it takes care of the input and output format
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
# plyr can easily be run in parallel -- faster
> strsplit(namedf, " ")
[[1]]
[1] "John" "ABC"


[[2]]
Tutorials
[1] "Mary" "CDE"
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.


[[3]]
Examples of using dplyr:
[1] "Kat" "FGH"
* [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 %>%.


> sapply(strsplit(namedf, " "), "[", 1)
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
[1] "John" "Mary" "Kat"
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"</syntaxhighlight>


=== Mean of duplicated columns: rowMeans; compute Means by each row ===
Tibble objects
* [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'>
* it does not have row names (cf data frame),
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
* it never changes the type of the inputs (e.g. it never converts strings to factors!),  
colnames(x) <- c("A","A", "b", "b", "b", "c"); x
* it never changes the names of variables
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res


# vapply() is safter than sapply().
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
# 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)
{{Pre}}
# [1] 31 27 28 29 30 31 32 33 34 35
> data(pew, package = "efficient")
</syntaxhighlight>
> dim(pew)  
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"    "tbl"        "data.frame"


=== Mean of duplicated rows: colMeans and rowsum ===
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
* colMeans(x, na.rm = FALSE, dims = 1)
# A tibble: 162 x 3
: <syntaxhighlight lang='rsplus'>
                                                      religion Income Count
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
                                                          <chr>  <chr> <int>
rownames(x) <- c(rep("a", 2), rep("b", 3), rep("c", 4), "d")
1                                                    Agnostic  <$10k    27
res <- sapply(split(1:nrow(x), rownames(x)),
2                                                      Atheist  <$10k    12
              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
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
</syntaxhighlight>
[1] NA
* rowsum(x, group, reorder = TRUE, …)
Warning message:
: <syntaxhighlight lang='rsplus'>
In mean.default(tidyr::gather(pew, key = Income, value = Count, :
x <- matrix(runif(100), ncol = 5) # 20 x 5
  argument is not numeric or logical: returning NA
group <- sample(1:8, 20, TRUE)
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
(xsum <- rowsum(x, group)) # 8 x 5
[1] 181.6975
</syntaxhighlight>
</pre>
* [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)
To show all rows of a tibble object, use the '''print()''' method.
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
<pre>
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
print(tbObj, n= Inf)
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
 
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
tbObj %>% print(n= nrow(.))
* [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'>
</pre>
> attach(mtcars)
 
dim(mtcars)
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.
[1] 32 11
 
> head(mtcars)
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].  
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
{{Pre}}
Mazda RX4        21.0  6  160 110 3.90 2.620 16.46  0  1    4    4
TibbleObject$VarName
Mazda RX4 Wag    21.0  6  160 110 3.90 2.875 17.02  0  1    4    4
# OR
Datsun 710        22.8  4  108  93 3.85 2.320 18.61  1  1    4    1
TibbleObject[["VarName"]]
Hornet 4 Drive    21.4  6  258 110 3.08 3.215 19.44  1  0    3    1
# OR
Hornet Sportabout 18.7  8  360 175 3.15 3.440 17.02  0  0    3    2
pull(TibbleObject, VarName) # won't be a tibble object anymore
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)
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
> mydf <- read.table(header=T, text='
# OR
id xval yval
dplyr::select(TibbleObject, 2:5) #
A 1  1
</pre>
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 ==
'''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]
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.
<pre>
my_data <- as_tibble(iris)
class(my_data)
</pre>
 
To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)


The following list gives a hierarchical relationship among these functions.
=== llply() ===
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
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.
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
<pre>
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
*** '''replicate'''(n, expr, simplify = "array")
</pre>
** '''mapply'''(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
where rLLID is a list of entrez ID. For example,
*** '''Vectorize'''(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
<pre>
*** '''Map'''(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
get("org.Hs.egGO")[["6772"]]
** '''vapply'''(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
</pre>
** '''rapply'''(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
returns a list of 49 GOs.
* '''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?]
=== ddply() ===
* 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
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
* 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:
=== ldply() ===
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
[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]
* [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 ===
=== Performance/speed comparison ===
Note that, apply's performance is not always better than a for loop. See
[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]
* 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`?]


=== Progress bar ===
== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
[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?]
http://rorynolan.rbind.io/2018/09/30/rcsetseed/


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.
=== get_seed() ===
See the same blog
{{Pre}}
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
=== Random seeds ===
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
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].  
* 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>
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming
set.seed(as.numeric(Sys.time()))
* 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)
set.seed(as.numeric(Sys.Date()))  # same seed for each day
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
</pre>
Map(weighted.mean, xs, ws)


# instead of a more clumsy way
=== .Machine and the largest integer, double ===
lapply(seq_along(xs), function(i) {
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].
  weighted.mean(xs[[i]], ws[[i]])
{{Pre}}
})
                          Linux/Mac  32-bit Windows 64-bit Windows
</syntaxhighlight>
double.eps              2.220446e-16  2.220446e-16  2.220446e-16
** 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'>
double.neg.eps          1.110223e-16  1.110223e-16  1.110223e-16
# Syntax: Reduce(f, x, ...)
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
 
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
> m1 <- data.frame(id=letters[1:4], val=1:4)
double.base            2.000000e+00  2.000000e+00  2.000000e+00
> m2 <- data.frame(id=letters[2:6], val=2:6)
double.digits          5.300000e+01  5.300000e+01  5.300000e+01
> merge(m1, m2, "id", all = T)
double.rounding        5.000000e+00   5.000000e+00  5.000000e+00
   id val.x val.y
double.guard            0.000000e+00  0.000000e+00  0.000000e+00
1  a    1    NA
double.ulp.digits      -5.200000e+01 -5.200000e+01 -5.200000e+01
2 b    2    2
double.neg.ulp.digits -5.300000e+01 -5.300000e+01 -5.300000e+01
3 c    3    3
double.exponent        1.100000e+01   1.100000e+01  1.100000e+01
4 d    4    4
double.min.exp        -1.022000e+03  -1.022000e+03 -1.022000e+03
e    NA    5
double.max.exp          1.024000e+03  1.024000e+03  1.024000e+03
6 f    NA    6
integer.max            2.147484e+09  2.147484e+09  2.147484e+09
> m <- list(m1, m2)
sizeof.long            8.000000e+00  4.000000e+00  4.000000e+00
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
sizeof.longlong        8.000000e+00  8.000000e+00  8.000000e+00
   id val.x val.y
sizeof.longdouble      1.600000e+01  1.200000e+01  1.600000e+01
a    1   NA
sizeof.pointer          8.000000e+00  4.000000e+00  8.000000e+00
2 b    2     2
</pre>
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 ===
=== NA when overflow ===
* [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).
tmp <- 156287L
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
tmp*tmp
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647
</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].
== How to select a seed for simulation or randomization ==
* [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 ]


=== rapply - recursive version of lapply ===
== set.seed() allow alphanumeric seeds ==
* http://4dpiecharts.com/tag/recursive/
https://stackoverflow.com/a/10913336
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


=== replicate ===
== set.seed(), for loop and saving random seeds ==
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
<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]
> replicate(5, rnorm(3))
<pre>
          [,1]       [,2]       [,3]      [,4]        [,5]
if (interactive()) {
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
  invisible(addTaskCallback(local({
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
    last <- .GlobalEnv$.Random.seed
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
   
</syntaxhighlight>
    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!


See [[#parallel_package|parSapply()]] for a parallel version of replicate().
.Random.seed <- seeds[[23]]  # restore
data.23 <- runif(5)  
data.23
data[[23]]  
</pre>
</li>
</ul>
* [https://www.rdocumentation.org/packages/impute/versions/1.46.0/topics/impute.knn impute.knn]  
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].


=== Vectorize ===
== sample() ==
* [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'>
=== sample() inaccurate on very large populations, fixed in R 3.6.0 ===
> rep(1:4, 4:1)
* [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.  
[1] 1 1 1 1 2 2 2 3 3 4
{{Pre}}
> vrep <- Vectorize(rep.int)
# R 3.5.3
> vrep(1:4, 4:1)
set.seed(123)
[[1]]
m <- (2/5)*2^32
[1] 1 1 1 1
m > 2^31
 
# [1] FALSE
[[2]]
log10(m)
[1] 2 2 2
# [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


[[3]]
# R 3.6.0
[1] 3 3
# 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
</pre>


[[4]]
=== Getting different results with set.seed() in RStudio ===
[1] 4
[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().''
</syntaxhighlight>
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy] <syntaxhighlight lang='rsplus'>
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
[1] 1.6491761 0.9610109
</syntaxhighlight>
* https://blogs.msdn.microsoft.com/gpalem/2013/03/28/make-vectorize-your-friend-in-r<syntaxhighlight lang='rsplus'>
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15


myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
=== dplyr::sample_n() ===
myfunc2(1, 2) # 2
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.
myfunc2(3, 5) # 15
myfunc2(c(1,3), c(2,5)) # NA
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used
</syntaxhighlight>


== plyr and dplyr packages ==
== Regular Expression ==
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
See [[Regular_expression|here]].


[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
== 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/


[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.
== 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.
<ul>
<li>Database connections
<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)
}
</pre>
</ul>


# plyr has a common syntax -- easier to remember
== file, connection ==
# plyr requires less code since it takes care of the input and output format
* [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)
# plyr can easily be run in parallel -- faster
* read() and write()
* read.table() and write.table()
{{Pre}}
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)


Tutorials
foo <- function() {
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
  con <- file()
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
  ...
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.
  on.exit(close(con))
  ...
}
</pre>
[https://r.789695.n4.nabble.com/Why-I-get-this-error-Error-in-close-connection-f-invalid-connection-td904413.html Error in close.connection(f) : invalid connection]. If we want to use '''close(con)''', we have to specify how to '''open''' the connection; such as
<pre>
con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)
</pre>


Examples of using dplyr:
=== withr package ===
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.  


=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
== Clipboard (?connections), textConnection(), pipe() ==
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
<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>


Tibble objects
=== clipr ===
* it does not have row names (cf data frame),
[https://cran.rstudio.com/web/packages/clipr/ clipr]: Read and Write from the System Clipboard
* 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]
== read/manipulate binary data ==
* x <- readBin(fn, raw(), file.info(fn)$size)
* rawToChar(x[1:16])
* See Biostrings C API


<syntaxhighlight lang='rsplus'>
== String Manipulation ==
> data(pew, package = "efficient")
* [https://www.gastonsanchez.com/r4strings/ Handling Strings with R](ebook) by Gaston Sanchez.
> dim(pew)  
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
[1] 18 10
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
> class(pew) # tibble is also a data frame!!
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
[1] "tbl_df"    "tbl"        "data.frame"
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]
* [https://statisticaloddsandends.wordpress.com/2022/05/05/a-short-note-on-the-startswith-function/ A short note on the startsWith function]


> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
=== format(): padding with zero ===
# A tibble: 162 x 3
<pre>
                                                      religion Income Count
ngenes <- 10
                                                          <chr>  <chr> <int>
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
1                                                    Agnostic  <$10k    27
[1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
2                                                      Atheist  <$10k    12
</pre>
...
> 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
</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.


'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].  
=== noquote() ===
<syntaxhighlight lang='rsplus'>
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.
TibbleObject$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
=== stringr package ===
# OR
* https://stringr.tidyverse.org/index.html
dplyr::select(TibbleObject, 2:5) #
* [https://stringr.tidyverse.org/articles/from-base.html Vignette compares stringr functions to their base R equivalents]
</syntaxhighlight>
* 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].


=== llply() ===
=== glue package ===
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.
<ul>
<li>[https://cran.r-project.org/web/packages/glue/index.html glue]. Useful in a loop and some function like ggtitle() or ggsave().
<pre>
<pre>
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
library(glue)
name <- "Fred"
glue('My name is {name}.') # My name is Fred.
</pre>
</pre>
where rLLID is a list of entrez ID. For example,  
</li>
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
</ul>
 
=== Raw data type ===
[https://twitter.com/hadleywickham/status/1387747735441395712 Fun with strings], [https://en.wikipedia.org/wiki/Cyrillic_alphabets Cyrillic alphabets]
<pre>
<pre>
get("org.Hs.egGO")[["6772"]]
a1 <- "А"
</pre>  
a2 <- "A"
returns a list of 49 GOs.
a1 == a2
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41
</pre>


=== ddply() ===
=== number of characters limit ===
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]


=== ldply() ===
== HTTPs connection ==  
[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]
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.


== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
<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.


=== get_seed() ===
The following R command will show the exact svn revision for the R you are currently using.
See the same blog
<pre>
<syntaxhighlight lang='rsplus'>
R.Version()$"svn rev"
get_seed <- function() {
</pre>
  sample.int(.Machine$integer.max, 1)
}
</syntaxhighlight>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== .Machine and the largest integer, double ===
If setInternet2(T), then https protocol is supported in download.file().  
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].
<syntaxhighlight lang='rsplus'>
                          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
</syntaxhighlight>


== How to select a seed for simulation or randomization ==
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.  
* [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 ]


== set.seed() allow alphanumeric seeds ==
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].
https://stackoverflow.com/a/10913336
 
== set.seed(), for loop and saving random seeds ==
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.


'''R up to 3.2.2'''
<pre>
<pre>
set.seed(1001)
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
data <- vector("list", 30)  
seeds <- vector("list", 30)
for(i in 1:30) {
  seeds[[i]] <- .Random.seed
  data[[i]] <- runif(5)  
}
.Random.seed <- seeds[[23]]  # restore
data.23 <- runif(5)  
data.23
data[[23]]
</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.seedSee ?.Random.seed for details.''
See also
* 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.''
* <src/include/Internal.h> (declare do_setInternet2()),  
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
* <src/main/names.c> (show do_setInternet2() in C)
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].
* <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).


== sample() inaccurate on very large populations, fixed in R 3.6.0 ==
'''R 3.3.0'''
* [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>
# R 3.5.3
setInternet2 <- function(use = TRUE) {
set.seed(123)
    if(!is.na(use)) stop("use != NA is defunct")
m <- (2/5)*2^32
    NA
m > 2^31
}
# [1] FALSE
</pre>
log10(m)
 
# [1] 9.23502
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.
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1
# 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
== Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan() ==
# docker run --net=host -it --rm r-base:3.6.0
In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, '''Inf''' and '''NaN''' as input or output.
> 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>
See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].


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


== Read rrd file ==
== replace() function ==
* https://en.wikipedia.org/wiki/RRDtool
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)
* http://oss.oetiker.ch/rrdtool/
* https://stackoverflow.com/a/11811147
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/


== file, connection ==
== File/path operations ==
* [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)
* list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
* read() and write()
* file.info()
* read.table() and write.table()
* dir.create()
<syntaxhighlight lang='rsplus'>
* file.create()
out = file('tmp.txt', 'w')
* file.copy()
writeLines("abcd", out)
* file.exists()
writeLines("eeeeee", out)
<ul>
close(out)
<li>'''basename'''() - remove the parent path, '''dirname'''() - returns the part of the path up to but excluding the last path separator
readLines('tmp.txt')
<pre>
unlink('tmp.txt')
> file.path("~", "Downloads")
args(writeLines)
[1] "~/Downloads"
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)
> dirname(file.path("~", "Downloads"))
</syntaxhighlight>
[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.


== Clipboard (?connections), textConnection(), pipe() ==
== read/download/source a file from internet ==
* On Windows, we can use readClipboard() and writeClipboard(). <syntaxhighlight lang='rsplus'>
=== Simple text file http ===
source("clipboard")
<pre>
read.table("clipboard")
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
</syntaxhighlight>
</pre>
* 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 ==
=== Zip, RData, gz file and url() function ===
* x <- readBin(fn, raw(), file.info(fn)$size)
<pre>
* rawToChar(x[1:16])
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
* See Biostrings C API
</pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.


== String Manipulation ==
Another example is [https://stackoverflow.com/a/9548672 Read gzipped csv directly from a url in R]
* [http://gastonsanchez.com/blog/resources/how-to/2013/09/22/Handling-and-Processing-Strings-in-R.html ebook] by Gaston Sanchez.
<pre>
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
                      "aa_combined-20110321.csv.gz", sep="")))
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
txt <- readLines(con)
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
dat <- read.csv(textConnection(txt))
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
</pre>
* [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 ===
Another example of using url() is
<pre>
<pre>
ngenes <- 10
load(url("http:/www.example.com/example.RData"))
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
</pre>
</pre>


== HTTPs connection ==
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].
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)
'''Dropbox''' is easy and works for load(), wget, ...


== setInternet2 ==
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.
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.
=== zip function ===
This will include 'hallmarkFiles' root folder in the files inside zip.
<pre>
<pre>
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
zip(zipfile = 'myFile.zip',  
              "GCF_000001405.13.assembly.txt")
    files = dir('hallmarkFiles', full.names = TRUE))
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.
# Verify/view the files. 'list = TRUE' won't extract
<pre>
unzip('testZip.zip', list = TRUE)  
R.Version()$"svn rev"
</pre>
</pre>


If setInternet2(T), then https protocol is supported in download.file().  
=== [http://cran.r-project.org/web/packages/downloader/index.html 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.


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.
=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
{{Pre}}
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))
</pre>


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].
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]


'''R up to 3.2.2'''
=== Github files https using RCurl package ===
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
<pre>
<pre>
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
</pre>
See also
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package
* <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).
== 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.


'''R 3.3.0'''
=== skimr: A frictionless, pipeable approach to dealing with summary statistics ===
<pre>
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]
setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>


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.
=== 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


== File operation ==
=== broom ===
* list.files()
[[Tidyverse#broom|Tidyverse->broom]]
* file.info()
* dir.create()
* file.create()
* file.copy()
* file.exists()
* basename() - remove the parent path
* 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 ==
=== Create publication tables using '''tables''' package ===
=== Simple text file http ===
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]
<pre>
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
</pre>


=== Zip file and url() function ===
R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,  
<pre>
{{Pre}}
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.
 
Another example of using url() is
<pre>
load(url("http:/www.example.com/example.RData"))
</pre>
 
=== [http://cran.r-project.org/web/packages/downloader/index.html 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 [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
<pre>
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))
</pre>
 
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]
 
=== Github files https using RCurl package ===
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
<pre>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package
 
== 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 ==
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]
 
== Create publication tables using '''tables''' package ==
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,  
<syntaxhighlight lang='rsplus'>
> library(tables)
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
Line 3,759: 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 3,777: 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 ==
== 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 ===
=== tableone ===
https://cran.r-project.org/web/packages/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 ===
=== dplyr ===
Line 3,798: Line 4,133:


=== base::prop.table(x, margin) ===
=== 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>
<pre>
R> m <- matrix(1:4, 2)
R> m <- matrix(1:4, 2)
Line 3,813: Line 4,149:


=== stats::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 3,902: 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 3,954: 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 3,981: 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 3,992: 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 3,998: 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
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)


=== Create a factor from a continuous variable: cut() ===
data.frame(sc=x, bins=bins) %>%
<syntaxhighlight lang='rsplus'>
  group_by(bins) %>%
facVar <- cut(contVar, c(-3, 0, 2, 4), labels = c("low", "medium", "high"))
  summarise(n=n()) %>%
</syntaxhighlight>
  ggplot(aes(x = bins, y = n)) +
Or [https://youtu.be/7oyiPBjLAWY?t=2480 a tibble object]
    geom_col(color = "black", fill = "#90AACB") +
<syntaxhighlight lang='rsplus'>
    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 4,018: 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>
 
=== How to change one of the level to NA ===
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
<pre>
x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA
</pre>
 
[https://webbedfeet.netlify.app/post/creating-missing-values-in-factors/ Creating missing values in factors]
 
=== Concatenating two factor vectors ===
Not trivial. [https://stackoverflow.com/a/5068939 How to concatenate factors, without them being converted to integer level?].
<pre>
unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
</pre>
 
=== droplevels() ===
[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.
 
=== factor(x , levels = ...) vs levels(x) <-  ===
<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.
 
{| class="wikitable"
|-
| [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
|}
 
<syntaxhighlight lang='rsplus'>
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
</syntaxhighlight>
A regression example.
<syntaxhighlight lang='rsplus'>
set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  0.96804    0.06610  14.65  <2e-16 ***
# x2          0.99620    0.09462  10.53  <2e-16 ***
 
# Wrong way when we want to change the baseline level to '2'
# No change on the model fitting except the apparent change on the variable name in the printout
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  0.96804    0.06610  14.65  <2e-16 ***
# x1          0.99620    0.09462  10.53  <2e-16 ***
 
# Correct way if we want to change the baseline level to '2'
# The estimate was changed by flipping the sign from the original data
sample_data$x <- relevel(x, ref = "2")
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  1.96425    0.06770  29.01  <2e-16 ***
# x1          -0.99620    0.09462  -10.53  <2e-16 ***
</syntaxhighlight>
 
=== stats::relevel() ===
[https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/relevel relevel]. This function can only be used to change the '''reference level''' of a factor variable. '''It does not directly create an arbitrary order of levels'''. That is, it is useful in lm() or aov(), etc.
 
=== reorder(), levels() and boxplot() ===
<ul>
<li>[https://www.r-bloggers.com/2023/09/how-to-reorder-boxplots-in-r-a-comprehensive-guide/ How to Reorder Boxplots in R: A Comprehensive Guide] (tapply() method, simple & effective)
<li>[https://stat.ethz.ch/R-manual/R-devel/library/stats/html/reorder.factor.html reorder()].This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
<pre>
# Syntax:
# newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
 
(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
</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
 
ordered(levels=c("a", "b", "c"))
# Error in factor(x, ..., ordered = TRUE) :
#  argument "x" is missing, with no default
</pre>


== Data frame ==
== Data frame ==
* http://adv-r.had.co.nz/Data-structures.html#data-frames
* 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/
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/


=== stringsAsFactors = FALSE ===
=== stringsAsFactors = FALSE ===
http://www.win-vector.com/blog/2018/03/r-tip-use-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, [https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/ stringAsFactors=FALSE] will be default. This also affects read.table() function.


=== check.names = FALSE ===
=== 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.
<pre>
<pre>
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
Line 4,037: Line 4,590:
1  1  1
1  1  1
2  2  2
2  2  2
</pre>
=== Create unique rownames: make.unique() ===
<pre>
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
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>
</pre>


Line 4,050: Line 4,619:
=== Convert data frame factor columns to characters ===
=== Convert data frame factor columns to characters ===
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
<syntaxhighlight lang='rsplus'>
{{Pre}}
# Method 1:
# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
Line 4,056: Line 4,625:
# Method 2:
# Method 2:
bob[] <- lapply(bob, as.character)
bob[] <- lapply(bob, as.character)
</syntaxhighlight>
</pre>
 
[https://stackoverflow.com/a/2853231 To replace only factor columns]:
<pre>
# 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
</pre>
 
=== Sort Or Order A Data Frame ===
[https://howtoprogram.xyz/2018/01/07/r-how-to-order-a-data-frame/ How To Sort Or Order A Data Frame In R]
# df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
# library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
# library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
# library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)


=== data.frame to vector ===
=== data.frame to vector ===
<syntaxhighlight lang='rsplus'>
<pre>
> a= matrix(1:6, 2,3)
df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))
> rownames(a) <- c("a", "b")
 
> colnames(a) <- c("x", "y", "z")
class(df)
> a
# [1] "data.frame"
  x y z
class(t(df))
a 1 3 5
# [1] "matrix" "array"
b 2 4 6
class(unlist(df))
> unlist(data.frame(a))
# [1] "numeric"
x1 x2 y1 y2 z1 z2
 
  1  2  3  4  5  6  
# Method 1: Convert data frame to matrix using as.matrix()
</syntaxhighlight>
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)  # [1] 1 2 3 4 5 6
vec2 <- c(mat)
 
# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))
 
# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6
 
# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3
# 1  2  3  4  5  6
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6
</pre>
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 [http://adv-r.had.co.nz/Data-structures.html#data-frames Advanced R -> Data structures] chapter.
 
=== cbind NULL and data.frame ===
[https://9to5tutorial.com/cbind-can-t-combine-null-with-dataframe cbind can't combine NULL with dataframe]. Add as.matrix() will fix the problem.


=== merge ===
=== merge ===
[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://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>
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>
 
=== is.matrix: data.frame is not necessarily a matrix ===
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/matrix ?matrix]. is.matrix returns TRUE '''if x is a vector and has a "dim" attribute of length 2''' and FALSE otherwise.
 
An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by
<pre>
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.
 
Another example that is a data frame but not a matrix is the built-in object ''cars''; see ?matrix. It is not a vector


=== matrix vs data.frame ===
=== Convert a data frame to a matrix: as.matrix() vs data.matrix() ===
<syntaxhighlight lang='rsplus'>
If I have a data frame X which recorded the time of some files.
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
 
unique(ip1$Priority)
* is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
# Error in ip1$Priority : $ operator is invalid for atomic vectors
* as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
unique(ip1[, "Priority"])   # OK
* [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.


ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
<syntaxhighlight lang='r'>
unique(ip2$Priority)     # OK
# 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
</syntaxhighlight>
</syntaxhighlight>


The length of a matrix and a data frame is different.
* 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='rsplus'>
* 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.
<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>
 
=== matrix vs data.frame ===
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
<pre>
Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
[1] "time"  "status" "treat"  "AKT1"  "BRAF"  "FLOT2"  "MTOR"  "PCK2"  "PIK3CA"
[10] "RAF1" 
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10
</pre>
 
Case 2:
{{Pre}}
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])  # OK
 
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
unique(ip2$Priority)    # OK
</pre>
 
The length of a matrix and a data frame is different.
{{Pre}}
> length(matrix(1:6, 3, 2))
> length(matrix(1:6, 3, 2))
[1] 6
[1] 6
Line 4,102: Line 4,827:
> x[[1]]
> x[[1]]
[1] 1 2 3 4 5 6
[1] 1 2 3 4 5 6
</syntaxhighlight>
</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.
So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.
=== How to Remove Duplicates ===
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]


=== Convert a matrix (not data frame) of characters to numeric ===
=== Convert a matrix (not data frame) of characters to numeric ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
<syntaxhighlight lang='rsplus'>
{{Pre}}
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
     a    b
     a    b
Line 4,121: Line 4,849:
> sum(tmp)
> sum(tmp)
[1] 1.917
[1] 1.917
</syntaxhighlight>
</pre>
 
=== Convert Data Frame Row to Vector ===
as.numeric() or '''c()'''
 
=== Convert characters to integers ===
mode(x) <- "integer"
 
=== Non-Standard Evaluation ===
[https://thomasadventure.blog/posts/understanding-nse-part1/ Understanding Non-Standard Evaluation. Part 1: The Basics]


=== Select Data Frame Columns in R ===
=== Select Data Frame Columns in R ===
Line 4,130: Line 4,867:
* select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
* 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
* 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.
<pre>
class(USArrests)  # "data.frame"
USArrests$"Assault"
</pre>
Note that for both data frame and matrix objects, we need to use the '''[''' operator to extract columns and/or rows.
<pre>
USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#        Murder Assault
# Alabama  13.2    236
# Alaska    10.0    263
USArrests[c("Murder", "Assault")]  # all rows
tmp <- data(package="datasets")
class(tmp$results)  # "matrix" "array"
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix
</pre>
Note for a '''data.table''' object, we can extract columns using the column names without double quotes.
<pre>
data.table(USArrests)[1:2, list(Murder, Assault)]
</pre>
=== Add columns to a data frame ===
[https://datasciencetut.com/how-to-add-columns-to-a-data-frame-in-r/ How to add columns to a data frame in R]
=== Exclude/drop/remove data frame columns ===
* [https://datasciencetut.com/remove-columns-from-a-data-frame/ How to Remove Columns from a data frame in R]
* [https://www.listendata.com/2015/06/r-keep-drop-columns-from-data-frame.html R: keep / drop columns from data frame]
<pre>
# 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))]
</pre>
=== 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]


=== Danger of selecting rows from a data frame ===
=== Danger of selecting rows from a data frame ===
Line 4,150: Line 4,934:
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]
[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 ===
=== Create an empty data.frame ===
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
https://stackoverflow.com/questions/10689055/create-an-empty-data-frame
* [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, ]).
<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)


My example:  
# similar to above
<syntaxhighlight lang='rsplus'>
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))
 
# different data type
a <- data.frame(x1 = character(),
                x2 = numeric(),
                x3 = factor(),
                stringsAsFactors = FALSE)
</pre>
 
=== Objects from subsetting a row in a data frame vs matrix ===
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
<ul>
<li>Subsetting creates repeated rows. This will create unexpected rownames.
<pre>
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
</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)
str(trainData)
# 'data.frame': 503 obs. of  500 variables:
# 'data.frame': 503 obs. of  500 variables:
Line 4,170: Line 4,988:
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
#  row names were found from a short variable and have been discarded
#  row names were found from a short variable and have been discarded
</syntaxhighlight>
</pre>


'trees' data from the 'datasets' package
'trees' data from the 'datasets' package
Line 4,215: Line 5,033:
# Volume 10.3  10.3
# Volume 10.3  10.3


# Method 3: convert a data frame to a matrix
is.matrix(trees)
is.matrix(trees)
# [1] FALSE
# [1] FALSE
Line 4,235: Line 5,054:
#    8.3  70.0  10.3
#    8.3  70.0  10.3
</pre>
</pre>
</li>
</ul>
=== Convert a list to data frame ===
[https://www.statology.org/convert-list-to-data-frame-r/ How to Convert a List to a Data Frame in R].
<pre>
# method 1
data.frame(t(sapply(my_list,c)))
# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)
# method 3
library(data.table)
rbindlist(my_list)
</pre>
=== tibble and data.table ===
* [[R#tibble | tibble]]
* [[Tidyverse#data.table|data.table]]
=== Clean  a dataset ===
[https://finnstats.com/index.php/2021/04/04/how-to-clean-the-datasets-in-r/ How to clean the datasets in R]


== matrix ==
== matrix ==
=== Define and subset a matrix ===
* [https://www.tutorialkart.com/r-tutorial/r-matrix/ Matrix in R]
** It is clear when a vector becomes a matrix the data is transformed column-wisely ('''byrow''' = FALSE, by default).
** When subsetting a matrix, it follows the format: '''X[rows, colums]''' or '''X[y-axis, x-axis]'''.
<pre>
data <- c(2, 4, 7, 5, 10, 1)
A <- matrix(data, ncol = 3)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7  10
# [2,]    4    5    1
A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]    7  10
</pre>
=== Prevent automatic conversion of single column to vector ===
use '''drop = FALSE''' such as mat[, 1, drop = FALSE].
=== complete.cases(): remove rows with missing in any column ===
It works on a sequence of vectors, matrices and data frames.
=== NROW vs nrow ===
=== NROW vs nrow ===
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/nrow ?nrow]. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/nrow ?nrow]. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.


=== matrix (column-major order) multiply a vector ===
=== matrix (column-major order) multiply a vector ===
* [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) is following the column-major order]
* 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.


<syntaxhighlight lang='rsplus'>
{{Pre}}
> matrix(1:6, 3,2)
> matrix(1:6, 3,2)
     [,1] [,2]
     [,1] [,2]
Line 4,249: Line 5,117:
[2,]    2    5
[2,]    2    5
[3,]    3    6
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3)
> 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] [,2]
[1,]    1    4
[1,]    1    4
[2,]    4  10
[2,]    4  10
[3,]    9  18
[3,]    9  18
> matrix(1:6, 3,2) * c(1,2,3,4)
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
     [,1] [,2]
     [,1] [,2]
[1,]    1  16
[1,]    1  16
[2,]    4    5
[2,]    4    5
[3,]    9  12
[3,]    9  12
</syntaxhighlight>
</pre>
 
* [https://stackoverflow.com/a/20596490 How to divide each row of a matrix by elements of a vector in R]
 
=== add a vector to all rows of a matrix ===
[https://stackoverflow.com/a/39443126 add a vector to all rows of a matrix]. sweep() or rep() is the best.
 
=== sparse matrix ===
[https://stackoverflow.com/a/10555270 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 ==
* [https://statisticaloddsandends.wordpress.com/2020/10/19/attributes-in-r/ Attributes in R]
* [http://adv-r.had.co.nz/Data-structures.html#attributes Data structures] in "Advanced R"


== Print a vector by suppressing names ==
== Names ==
Use '''unname'''.
[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()


== format.pval ==
=== Print a vector by suppressing names ===
<syntaxhighlight lang='rsplus'>
Use '''unname'''. sapply(, , USE.NAMES = FALSE).
> args(format.pval)
function (pv, digits = max(1L, getOption("digits") - 2L), eps = .Machine$double.eps,  
    na.form = "NA", ...)


== format.pval/print p-values/format p values ==
[https://rdrr.io/r/base/format.pval.html format.pval()]. By default it will show 5 significant digits (getOption("digits")-2).
{{Pre}}
> format.pval(c(stats::runif(5), pi^-100, NA))
> format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"     
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"     
> format.pval(c(0.1, 0.0001, 1e-27))
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"
[1] "1e-01"  "1e-04"  "<2e-16"
</syntaxhighlight>
 
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
</pre>


== Customize R: options() ==
== Customize R: options() ==


=== Change the default R repository ===
=== Change the default R repository, my .Rprofile ===
[[Rstudio#Change_repository|Change R repository]]
[[Rstudio#Change_repository|Change R repository]]


Line 4,284: Line 5,179:


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'''.
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'''.
<syntaxhighlight lang='rsplus'>
 
Type '''file.edit("~/.Rprofile")'''
{{Pre}}
local({
local({
   r = getOption("repos")
   r = getOption("repos")
Line 4,290: Line 5,187:
   options(repos = r)
   options(repos = r)
})
})
options(continue = "  ")
options(continue = "  ", editor = "nano")
message("Hi MC, loading ~/.Rprofile")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
if (interactive()) {
   .Last <- function() try(savehistory("~/.Rhistory"))
   .Last <- function() try(savehistory("~/.Rhistory"))
}
}
</syntaxhighlight>
</pre>


=== Change the default web browser ===
=== Change the default web browser for utils::browseURL() ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey).
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put
<syntaxhighlight lang='rsplus'>
<pre>
> help.start()
options(browser='seamonkey')
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
<pre>
options(browser='seamonkey')
</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.
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:
For one-time only purpose, we can use the ''browser'' option in help.start() function:
<syntaxhighlight lang='rsplus'>
{{Pre}}
> help.start(browser="seamonkey")
> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
If the browser launched by 'seamonkey' is already running, it is *not*
     restarted, and you must switch to its window.
     restarted, and you must switch to its window.
Otherwise, be patient ...
Otherwise, be patient ...
</syntaxhighlight>
</pre>


We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See  
We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See  
Line 4,338: Line 5,215:
=== Change the default editor ===
=== Change the default editor ===
On my Linux and mac, the default editor is "vi". To change it to "nano",
On my Linux and mac, the default editor is "vi". To change it to "nano",
<syntaxhighlight lang='rsplus'>
{{Pre}}
options(editor = "nano")
options(editor = "nano")
</syntaxhighlight>
</pre>


=== Change prompt and remove '+' sign ===
=== Change prompt and remove '+' sign ===
See https://stackoverflow.com/a/1448823.
See https://stackoverflow.com/a/1448823.
<syntaxhighlight lang='rsplus'>
{{Pre}}
options(prompt="R> ", continue=" ")
options(prompt="R> ", continue=" ")
</syntaxhighlight>
</pre>


=== digits ===
=== digits ===
* [https://gist.github.com/arraytools/26a0b359541f4fc9fddc8f0a0c94489e Read and compute the sum of a numeric matrix file] using R vs Python vs C++. Note by default R does not show digits after the decimal point because the number is large.
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
* [https://stackoverflow.com/a/10712012 ?print.default]
* [https://stackoverflow.com/a/10712012 ?print.default]
* [https://stackoverflow.com/a/12135122 Formatting Decimal places in R, round()]. [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/format format()] where '''nsmall''' controls the minimum number of digits to the right of the decimal point
* [https://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
* [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


In R,
In R,
<syntaxhighlight lang='rsplus'>
{{Pre}}
> options()$digits # Default
> options()$digits # Default
[1] 7
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
> 100000.07 + .04
[1] 100000.1
[1] 100000.1
Line 4,363: Line 5,254:
> 100000.07 + .04
> 100000.07 + .04
[1] 100000.11
[1] 100000.11
</syntaxhighlight>
</pre>


In Python,
In Python,
<syntaxhighlight lang='python'>
{{Pre}}
>>> 100000.07 + .04
>>> 100000.07 + .04
100000.11
100000.11
</syntaxhighlight>
</pre>


=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
<syntaxhighlight lang='rsplus'>
[https://datasciencetut.com/how-to-turn-off-scientific-notation-in-r/ 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.
{{Pre}}
> numer = 29707; denom = 93874
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)  
> c(numer/denom, numer, denom)  
Line 4,389: Line 5,283:
> c(4/5, numer, denom)
> c(4/5, numer, denom)
[1]    0.8 29707.0 93874.0
[1]    0.8 29707.0 93874.0
</syntaxhighlight>
</pre>


=== Suppress warnings ===
=== Suppress warnings: options() and capture.output() ===
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.  
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.  
<syntaxhighlight lang='rsplus'>
{{Pre}}
op <- options("warn")
op <- options("warn")
options(warn = -1)
options(warn = -1)
Line 4,404: Line 5,298:
...
...
options(warn = warnLevel)
options(warn = warnLevel)
</syntaxhighlight>
</pre>
 
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/warning suppressWarnings()]
<pre>
suppressWarnings( foo() )
 
foo <- capture.output(
bar <- suppressWarnings(
{print( "hello, world" );
  warning("unwanted" )} ) )
</pre>
 
[https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/capture.output capture.output()]
<pre>
str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")
</pre>


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


== sprintf ==
== sprintf ==
=== paste, paste0, sprintf ===
=== paste, paste0, sprintf ===
[https://www.r-bloggers.com/paste-paste0-and-sprintf/ this post]
[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]
 
=== 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
<pre>
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"
</pre>


=== Format number as fixed width, with leading zeros ===
=== Format number as fixed width, with leading zeros ===
Line 4,417: Line 5,358:
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1


<syntaxhighlight lang='rsplus'>
{{Pre}}
# sprintf()
# sprintf()
a <- seq(1,101,25)
a <- seq(1,101,25)
Line 4,430: Line 5,371:
paste0("bm", gsub(" ", "0", format(5:15)))
paste0("bm", gsub(" ", "0", format(5:15)))
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
</syntaxhighlight>
</pre>
 
=== formatC and prettyNum (prettifying numbers) ===
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/formatC formatC() & prettyNum()]
* [[R#format.pval|format.pval()]]
<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"
 
R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"
</pre>


=== sprintf does not print ===
=== Format(x, scientific = TRUE) ===
Use cat() or print() outside sprintf(). sprintf() do not print in a non interactive mode.
Print numeric data in exponential format, so .0001 prints as 1e-4
<syntaxhighlight lang='rsplus'>
cat(sprintf('%5.2f\t%i\n',1.234, l234))
</syntaxhighlight>


== Creating publication quality graphs in R ==
== Creating publication quality graphs in R ==
Line 4,448: Line 5,415:
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
* [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.
* 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.
<syntaxhighlight lang='rsplus'>
> h5ls(destination_file)
  group                          name      otype  dclass          dim
0      /                          data  H5I_GROUP                     
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 ==
== Formats for writing/saving and sharing data ==
Line 4,506: Line 5,423:


== with() and within() functions ==
== with() and within() functions ==
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://www.r-bloggers.com/2023/07/simplify-your-code-with-rs-powerful-functions-with-and-within/ Simplify Your Code with R’s Powerful Functions: with() and within()]
* within() is similar to with() except it is used to create new columns and merge them with the original data sets. But if we just want to create a new column, we can just use df$newVar = . The following example is from [http://www.youtube.com/watch?v=pZ6Bnxg9E8w&list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP youtube video].
<pre>
<pre>
closePr <- with(mariokart, totalPr - shipPr)
closePr <- with(mariokart, totalPr - shipPr)
Line 4,525: Line 5,443:
</pre>
</pre>


== stem(): stem-and-leaf plot, bar chart on terminals ==
== stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals ==
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
* 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://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package
* [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>
== Histogram with density line ==
<pre>
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).


== Graphical Parameters, Axes and Text, Combining Plots ==
== Graphical Parameters, Axes and Text, Combining Plots ==
Line 4,534: Line 5,472:


== 15 Questions All R Users Have About Plots ==
== 15 Questions All R Users Have About Plots ==
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.
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.


# How To Draw An Empty R Plot? plot.new()
# How To Draw An Empty R Plot? plot.new()
# How To Set The Axis Labels And Title Of The R Plots?
# How To Set The Axis Labels And Title Of The R Plots?
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()  
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()  
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext()
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par].
# How To Add Or Change The R Plot’s Legend? legend()
# How To Add Or Change The R Plot’s Legend? legend()
# How To Draw A Grid In Your R Plot? grid()  
# How To Draw A Grid In Your R Plot? [https://r-charts.com/base-r/grid/ grid()]
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
# How To 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 Adjust The Size Of Points In An R Plot? cex argument
Line 4,547: Line 5,485:
# How To Add Error Bars In An R Plot? arrows()
# How To Add Error Bars In An R Plot? arrows()
# How To Save A Plot As An Image On Disc  
# 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 Two R Plots Next To Each Other? '''par(mfrow)'''[which means Multiple Figures (use ROW-wise)], '''gridBase''' package, '''lattice''' package
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
Line 4,554: Line 5,492:
== jitter function ==
== jitter function ==
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* [https://statistical-programming.com/jitter-r-function-example/ The jitter R Function] 3 Example Codes (Basic Application & Boxplot Visualization)
** jitter(, amount) function adds a random variation between -amount/2 and amount/2 to each element in x
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
* [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]
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]
:[[File:Jitterbox.png|200px]]


== Scatterplot with the "rug" function ==
== Scatterplot with the "rug" function ==
Line 4,572: Line 5,513:


== Identify/Locate Points in a Scatter Plot ==
== Identify/Locate Points in a Scatter Plot ==
[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
<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>


== Draw a single plot with two different y-axes ==
== Draw a single plot with two different y-axes ==
Line 4,580: Line 5,532:
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/


=== Default palette ===
=== Default palette before R 4.0 ===
palette() # black, red, green3, blue, cyan, magenta, yellow, gray
palette() # black, red, green3, blue, cyan, magenta, yellow, gray
<pre>
# 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)
</pre>
=== New palette in R 4.0.0 ===
[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")
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")
  }
</pre>
The '''palette''' function can also be used to change the color palette. See [https://data.library.virginia.edu/setting-up-color-palettes-in-r/ Setting up Color Palettes in R]
<pre>
palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
  # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))
cc <- palette()
palette(c(cc,"purple","brown")) # Add two colors
</pre>
<pre>
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502
</pre>
=== evoPalette ===
[http://gradientdescending.com/evolve-new-colour-palettes-in-r-with-evopalette/ Evolve new colour palettes in R with evoPalette]
=== rtist ===
[https://github.com/tomasokal/rtist?s=09 rtist]: Use the palettes of famous artists in your own visualizations.


== SVG ==
== SVG ==
Line 4,588: Line 5,597:


=== svglite ===
=== svglite ===
https://blog.rstudio.org/2016/11/14/svglite-1-2-0/
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].


=== pdf -> svg ===
=== pdf -> svg ===
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
=== svg -> png ===
[https://laustep.github.io/stlahblog/posts/SVG2PNG.html SVG to PNG] using the [https://cran.rstudio.com/web/packages/gyro/index.html gyro] package


== read.table ==
== read.table ==
=== clipboard ===
=== clipboard ===
<syntaxhighlight lang="rsplus">
{{Pre}}
source("clipboard")
source("clipboard")
read.table("clipboard")
read.table("clipboard")
</syntaxhighlight>
</pre>


=== inline text ===
=== inline text ===
<syntaxhighlight lang="rsplus">
{{Pre}}
mydf <- read.table(header=T, text='
mydf <- read.table(header=T, text='
  cond yval
  cond yval
Line 4,608: Line 5,621:
     C 1.6
     C 1.6
')
')
</syntaxhighlight>
</pre>


=== http(s) connection ===
=== http(s) connection ===
<syntaxhighlight lang="rsplus">
{{Pre}}
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
                           ssl.verifypeer = FALSE)
                           ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
ip <- read.table(textConnection(temp), as.is=TRUE)
</syntaxhighlight>
</pre>


=== read only specific columns ===
=== 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.
Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.
<syntaxhighlight lang="rsplus">
{{Pre}}
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),  
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),  
                 skip=62, header=T, stringsAsFactors = FALSE)[, 1]
                 skip=62, header=T, stringsAsFactors = FALSE)[, 1]
Line 4,625: Line 5,638:
system.time(x <- read.delim("Methylation450k.txt",  
system.time(x <- read.delim("Methylation450k.txt",  
                 colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
                 colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
</syntaxhighlight>
</pre>


To know the number of columns, we might want to read the first row first.
To know the number of columns, we might want to read the first row first.
<syntaxhighlight lang="rsplus">
{{Pre}}
library(magrittr)
library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</syntaxhighlight>
</pre>


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]
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]
=== 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" 
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" 
</pre>
=== setNames() ===
Change the colnames. See an example from [https://www.tidymodels.org/start/models/ tidymodels]
=== Testing for valid variable names ===
[https://www.r-bloggers.com/testing-for-valid-variable-names/ Testing for valid variable names]
=== make.names(): Make syntactically valid names out of character vectors ===
* [https://stat.ethz.ch/R-manual/R-devel/library/base/html/make.names.html make.names()]
* A valid variable name consists of letters, numbers and the '''dot''' or '''underline''' characters. The variable name starts with a letter or the dot not followed by a number. See [https://www.tutorialspoint.com/r/r_variables.htm R variables].
<pre>
make.names("abc-d") # [1] "abc.d"
</pre>


== Serialization ==
== Serialization ==
Line 4,747: Line 5,786:
=== Avoid Coercing Indices To Doubles ===
=== Avoid Coercing Indices To Doubles ===
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
=== Careful on NA value ===
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<pre>
R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3  3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6
</pre>
=== Implicit looping ===
<pre>
set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]    # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7
</pre>


== modelling ==
== modelling ==
Line 4,757: Line 5,827:


=== as.formula(): use a string in formula in lm(), glm(), ... ===
=== as.formula(): use a string in formula in lm(), glm(), ... ===
* [https://www.r-bloggers.com/2019/08/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
<syntaxhighlight lang='rsplus'>
{{Pre}}
? as.formula
? as.formula
xnam <- paste("x", 1:25, sep="")
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</syntaxhighlight>
</pre>
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
<syntaxhighlight lang='rsplus'>
{{Pre}}
outcome <- "mpg"
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")
variables <- c("cyl", "disp", "hp", "carb")
Line 4,812: Line 5,883:
(Intercept)          cyl        disp          hp        carb   
(Intercept)          cyl        disp          hp        carb   
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
</syntaxhighlight>
</pre>
* [https://statisticaloddsandends.wordpress.com/2019/08/24/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula] 1. as.formula() 2. subset by [[i]] 3. get() 4. eval(parse()).
* [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 ===
=== Aggregating results from linear model ===
https://stats.stackexchange.com/a/6862
https://stats.stackexchange.com/a/6862


== S3 and S4 methods ==
== Replacement function "fun(x) <- a" ==
[https://stackoverflow.com/questions/11563154/what-are-replacement-functions-in-r What are Replacement Functions in R?]
<pre>
R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
    x[x > value] <- Inf
    x
}
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]  1  3 Inf Inf
 
R> "cutoff<-"(x = xx, value = 65)
[1]  1  3 Inf Inf
</pre>
The statement '''fun(x) <- a''' and R will read '''x <- "fun<-"(x,a) '''
 
== S3 and S4 methods and signature ==
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* Software for Data Analysis: Programming with R by John Chambers
* Software for Data Analysis: Programming with R by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* [https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers, 2016
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods]
* [https://rstudio-education.github.io/hopr/s3.html Hands-On Programming with R] by Garrett Grolemund
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package]
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package], * [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://adv-r.had.co.nz/S4.html, http://adv-r.had.co.nz/OO-essentials.html
* http://adv-r.had.co.nz/S4.html
* [https://appsilon.com/object-oriented-programming-in-r-part-1/ Object-Oriented Programming in R (Part 1): An Introduction], [https://appsilon.com/object-oriented-programming-in-r-part-2/ Part 2: S3 Simplified]


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


* '''Debug a S4 function'''
<syntaxhighlight lang='rsplus'>
> library(genefilter) # Bioconductor
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
> showMethods("nsFilter")
Line 4,846: Line 5,941:
eset="ExpressionSet"
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")
> debug(nsFilter, signature="ExpressionSet")
</syntaxhighlight>
 
library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code
</pre>
See the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L367 source code] of '''normalizationFactors<-''' (setReplaceMethod() is used) and the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L385 source code] of '''estimateSizeFactors()'''. We can see how ''avgTxLength'' was used in estimateNormFactors().
 
Another example
<pre>
library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)
 
showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"
 
debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz,
#      normalization = ssgsea.norm, verbose = verbose,
#      BPPARAM = BPPARAM))
 
isdebugged("gsva")
# [1] TRUE
undebug(gsva)
</pre>


* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
<syntaxhighlight lang='rsplus'>
{{Pre}}
library(IRanges)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir <- IRanges(start=c(10, 20, 30), width=5)
Line 4,880: Line 6,009:
##  
##  
## Known Subclasses: "NormalIRanges"
## Known Subclasses: "NormalIRanges"
</syntaxhighlight>
</pre>
 
=== 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.
* [https://kasperdanielhansen.github.io/genbioconductor/html/R_S4.html#slots-and-accessor-functions R - S4 Classes and Methods] Hansen. '''getClass()''' or '''getClassDef()'''.
 
=== setReplaceMethod() ===
* [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?]


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


=== View S3 function definition: double colon '::' and triple colon ':::' operators ===
=== View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere() ===
?":::"
?":::"


Line 4,892: Line 6,039:
* pkg:::name returns the value of the internal variable name
* pkg:::name returns the value of the internal variable name


<syntaxhighlight lang='rsplus'>
<pre>
base::"+"
base::"+"
stats:::coef.default
stats:::coef.default
</syntaxhighlight>
 
predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR 
getS3method("predict", "ppr")
 
getS3method("t", "test")
</pre>
 
[https://stackoverflow.com/a/19226817 methods() + getAnywhere() functions]


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


=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
Line 4,907: Line 6,075:


For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
<syntaxhighlight lang='rsplus'>
{{Pre}}
> mcols(ddsNoPrior[genes, ])
> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
DataFrame with 2 rows and 21 columns
Line 4,926: Line 6,094:
1      TRUE        3  210.4045 0.2648753
1      TRUE        3  210.4045 0.2648753
2      TRUE        9  243.7455 0.3248949
2      TRUE        9  243.7455 0.3248949
</syntaxhighlight>
</pre>
 
== Pipe ==
* [https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
* [https://win-vector.com/2020/12/05/r-is-getting-an-official-pipe-operator/ R is Getting an Official Pipe Operator], [https://win-vector.com/2020/12/07/my-opinion-on-rs-upcoming-pipe/ My Opinion on R’s Upcoming Pipe]
<ul>
<li> a(b(x)) vs '''x |> b() |> a()'''. See [https://twitter.com/henrikbengtsson/status/1335328090390597632 this tweet] in R-dev 2020-12-04.
<pre>
e0 <- quote(a(b(x)))
e1 <- quote(x |> b() |> a())
identical(e0, e1)
</pre>
</li>
<li>
[https://selbydavid.com/2021/05/18/pipes/ There are now 3 different R pipes]
</li>
<li>[https://stackoverflow.com/a/67629310 Error: The pipe operator requires a function call as RHS].
<pre>
# native pipe
foo |> bar()
# magrittr pipe
foo %>% bar
</pre>
</li>
<li>[https://www.infoworld.com/article/3621369/use-the-new-r-pipe-built-into-r-41.html Use the new R pipe built into R 4.1] </li>
<li>[https://towardsdatascience.com/the-new-native-pipe-operator-in-r-cbc5fa8a37bd The New Native Pipe Operator in R] </li>
<li>[https://ivelasq.rbind.io/blog/understanding-the-r-pipe/ Understanding the native R pipe |> ] </li>
</ul>
 
Packages take advantage of pipes
<ul>
<li>[https://cran.r-project.org/web/packages/rstatix/index.html rstatix]: Pipe-Friendly Framework for Basic Statistical Tests
</ul>


== findInterval() ==
== findInterval() ==
Line 4,932: Line 6,132:
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
== Assign operator ==
* Earlier versions of R used underscore (_) as an assignment operator.
* [https://developer.r-project.org/equalAssign.html Assignments with the = Operator]
* In R 1.8.0 (2003), the assign operator has been removed. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
* In R 1.9.0 (2004), "_" is allowed in valid names. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
: [[File:R162.png|200px]]
== Operator precedence ==
The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.


== order(), rank() and sort() ==
== order(), rank() and sort() ==
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
<syntaxhighlight lang='rsplus'>
<pre>
> x = sample(10)
> x = sample(10)
> x
> x
Line 4,950: Line 6,161:
> sort(x)
> sort(x)
  [1]  1  2  3  4  5  6  7  8  9 10
  [1]  1  2  3  4  5  6  7  8  9 10
</syntaxhighlight>
</pre>
 
=== OS-dependent results on sorting string vector ===
Gene symbol case.
<pre>
# mac:
order(c("DC-UbP", "DC2")) # c(1,2)
 
# linux:
order(c("DC-UbP", "DC2")) # c(2,1)
</pre>
 
Affymetric id case.
<pre>
# mac:
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"


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


Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html. help].
The difference is related to locale. See


* Usage
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/locales ?locales] in R
: <syntaxhighlight lang='rsplus'>
* On OS, type '''locale'''
do.call(what, args, quote = FALSE, envir = parent.frame())
* [https://stackoverflow.com/questions/39171613/sort-produces-different-results-in-ubuntu-and-windows sort() produces different results in Ubuntu and Windows]
# what: either a function or a non-empty character string naming the function to be called.
* 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).
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
<pre>
# quote: a logical value indicating whether to quote the arguments.
# 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"))
</pre>
 
=== unique() ===
It seems it does not sort. [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/unique ?unique].
<pre>
# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"
</pre>
 
== 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.
 
[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]
 
Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html help].
 
* Usage
{{Pre}}
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
# 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.
#        if what is a character string and the arguments are symbols or quoted expressions.
</syntaxhighlight>
</pre>
* do.call() is similar to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/lapply lapply()] but not the same.
* 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.
: <syntaxhighlight lang='rsplus'>
{{Pre}}
> do.call("complex", list(imag = 1:3))
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
[1] 0+1i 0+2i 0+3i
Line 4,975: Line 6,237:
> complex(imag=1:3)
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
[1] 0+1i 0+2i 0+3i
</syntaxhighlight>
> 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>
* [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.]
* [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.]
:<syntaxhighlight lang='rsplus'>
{{Pre}}
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> length(tmp)
> length(tmp)
Line 5,005: Line 6,275:
  [1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
  [1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[12] "b3-"
[12] "b3-"
</syntaxhighlight>
</pre>
* ''environment'' and ''quote'' arguments.
* ''environment'' and ''quote'' arguments.
:<syntaxhighlight lang='rsplus'>
{{Pre}}
> A <- 2
> A <- 2
> f <- function(x) print(x^2)
> f <- function(x) print(x^2)
Line 5,031: Line 6,301:
> eval(call("f", quote(A)), envir = env)   
> eval(call("f", quote(A)), envir = env)   
[1] 100
[1] 100
</syntaxhighlight>
</pre>
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
:<syntaxhighlight lang='rsplus'>
{{Pre}}
> foo <- function(a=1, b=2, ...) {  
> foo <- function(a=1, b=2, ...) {  
         list(arg=do.call(c, as.list(match.call())[-1]))  
         list(arg=do.call(c, as.list(match.call())[-1]))  
Line 5,048: Line 6,318:
a b c  
a b c  
1 2 3  
1 2 3  
</syntaxhighlight>
</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>
 
=== expand.grid, mapply, vapply ===
[https://shikokuchuo.net/posts/10-combinations/ A faster way to generate combinations for mapply and vapply]


=== do.call vs mapply ===
=== 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.  
* 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'>
{{Pre}}
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
       Var1  
       Var1  
Line 5,082: Line 6,375:
[11,] "a"  "3"  "-"  
[11,] "a"  "3"  "-"  
[12,] "b"  "3"  "-"  
[12,] "b"  "3"  "-"  
</syntaxhighlight>
</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])
* 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])
:<syntaxhighlight lang='rsplus'>
{{Pre}}
set.seed(1)
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
Line 5,093: Line 6,386:
rowMeans(x)
rowMeans(x)
# [1]  1.032209 10.104131
# [1]  1.032209 10.104131
</syntaxhighlight>
</pre>
:<syntaxhighlight lang='rsplus'>
{{Pre}}
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))
</syntaxhighlight>
</pre>


=== do.call vs lapply ===
=== do.call vs lapply ===
https://stackoverflow.com/a/10801883
[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())'''
 
* 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.
* 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).'''
* 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.
* 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.


:<syntaxhighlight lang='rsplus'>
{{Pre}}
> lapply(iris, class) # same as Map(class, iris)
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
$Sepal.Length
Line 5,127: Line 6,421:
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species  
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species  
   "numeric"    "numeric"    "numeric"    "numeric"    "factor"  
   "numeric"    "numeric"    "numeric"    "numeric"    "factor"  
</syntaxhighlight>
</pre>


https://stackoverflow.com/a/10801902  
https://stackoverflow.com/a/10801902  
* '''lapply''' applies a function '''over a list'''. So there will be several function calls.
* '''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 [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.
* '''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.
: <syntaxhighlight lang='rsplus'>
{{Pre}}
> X <- list(1:3,4:6,7:9)
> X <- list(1:3,4:6,7:9)
> lapply(X,mean)
> lapply(X,mean)
Line 5,169: Line 6,463:
> mapply(mean, X)  
> mapply(mean, X)  
[1] 2 5 8
[1] 2 5 8
</syntaxhighlight>
</pre>
Below is a good example to show the difference of lapply() and do.call() - [https://stackoverflow.com/a/42734863 Generating Random Strings].  
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'>
{{Pre}}
> set.seed(1)
> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
Line 5,188: Line 6,482:
[1] "B" "W" "K" "N"
[1] "B" "W" "K" "N"


> lapply(x, paste0, collapse= '')
> lapply(x, paste0, collapse= "")
[[1]]
[[1]]
[1] "YDGA"
[1] "YDGA"
Line 5,197: Line 6,491:
> do.call(paste0, x)
> do.call(paste0, x)
[1] "YB" "DW" "GK" "AN"
[1] "YB" "DW" "GK" "AN"
</syntaxhighlight>
</pre>


=== do.call + rbind + lapply ===
=== 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.
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.
<syntaxhighlight lang='rsplus'>
{{Pre}}
x <- readLines(textConnection("---CLUSTER 1 ---
x <- readLines(textConnection("---CLUSTER 1 ---
  3
  3
Line 5,239: Line 6,533:
[7,] "8"    "2"
[7,] "8"    "2"
[8,] "11"  "2"
[8,] "11"  "2"
</syntaxhighlight>
</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()).
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().
Another example is to reproduce aggregate(). aggregate() = do.call() + by().
<syntaxhighlight lang='rsplus'>
{{Pre}}
attach(mtcars)
attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
Line 5,250: Line 6,544:
# except it does not have an extra Group.x columns
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)
aggregate(mtcars, list(cyl, vs), FUN=mean)
</syntaxhighlight>
</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>


== How to get examples from help file ==
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
Method 1:
Method 1:
Line 5,274: Line 6,585:
</pre>
</pre>


== Dealing with date ==
== Dealing with dates ==
<pre>
* 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()
d1 = date()
class(d1) # "character"
class(d1) # "character"
Line 5,293: Line 6,623:
wday(x[1]) # 3
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues
wday(x[1], label=TRUE) # Tues
</pre>
</syntaxhighlight>
* http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
* 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
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
* We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.
* 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
* [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'''.


== [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] and deparse/substitute ==
* [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
* [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
** Labelling: turn an argument into a label
** Formulas
** Formulas
** Dot-dot-dot
** Dot-dot-dot
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression.  
* [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.  
** 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.
** 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
* 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.
:<syntaxhighlight lang='rsplus'>
* deparse(expr) - turns unevaluated expressions into character strings. For example,  
mode(quote(1:10))
<pre>
# [1] "call"
> deparse(args(lm))
</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\", "  
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "  
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
Line 5,324: Line 6,709:
[7] "    contrasts = NULL, "          "    offset, ...) "               
[7] "    contrasts = NULL, "          "    offset, ...) "               
[9] "NULL"
[9] "NULL"
</pre>
</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!  
* 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]  
** [http://r.789695.n4.nabble.com/using-eval-parse-paste-in-a-loop-td849207.html eval(parse...)) should generally be avoided]  
Line 5,330: Line 6,715:


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).  
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).  
<syntaxhighlight lang='rsplus'>
{{Pre}}
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3


Line 5,352: Line 6,737:
})
})
names(res) <- method
names(res) <- method
</syntaxhighlight>
</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 ==
== 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
* 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 ==
== 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 ===
=== 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.
[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.
Line 5,367: Line 6,769:


Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
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 ===
=== Lazy evaluation in R functions arguments ===
* http://adv-r.had.co.nz/Functions.html
* http://adv-r.had.co.nz/Functions.html
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.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'''.  
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  
Line 5,422: Line 6,838:
=== Use of functions as arguments ===
=== Use of functions as arguments ===
[https://www.njtierney.com/post/2019/09/29/unexpected-function/ Just Quickly: The unexpected 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 ==  
== 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 this note].
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.
'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
Line 5,433: Line 6,882:
</pre>
</pre>


== List data type ==
Use with functions like sapply, e.g. '''sapply(1:5, `+`, 3) '''  .
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
 
== 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>
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
defaultW <- getOption("warn")  
> do.call(mean, args)
options(warn = -1)  
[1] 5.5
[YOUR CODE]  
> mean(c(1:10, NA, NA), na.rm = TRUE)
options(warn = defaultW)
[1] 5.5
</pre>
</pre>
 
</li>
== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
<li>try() allows execution to continue even after an error has occurred. You can suppress the message with '''try(..., silent = TRUE)'''.
* http://adv-r.had.co.nz/Exceptions-Debugging.html
* try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
<pre>
<pre>
out <- try({
out <- try({
Line 5,458: Line 6,914:
succeeded <- !sapply(results, is.error)
succeeded <- !sapply(results, is.error)
</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().
</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>
<pre>
tryCatch(expr, ..., finally)
tryCatch(expr, ..., finally)
Line 5,479: Line 6,936:
</pre>
</pre>
Below is another snippet from available.packages() function,
Below is another snippet from available.packages() function,
<syntaxhighlight lang='rsplus>
{{Pre}}
z <- tryCatch(download.file(....), error = identity)
z <- tryCatch(download.file(....), error = identity)
if (!inherits(z, "error")) STATEMENTS
if (!inherits(z, "error")) STATEMENTS
</syntaxhighlight>
</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].


== Using list type ==
=== Avoid if-else or switch ===
=== Avoid if-else or switch ===
?plot.stepfun.
?plot.stepfun.
Line 5,516: Line 7,020:
?par  
?par  


=== text size and font on main, lab & axis ===
=== text size (cex) and font size on main, lab & axis ===
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
* [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]
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]


Examples:
Examples (default is 1 for each of them):
* cex.main=0.9
* cex.main=0.9  
* cex.lab=0.8
* cex.sub
* font.lab=2
* cex.lab=0.8, font.lab=2 (x/y axis labels)
* cex.axis=0.8
* cex.axis=0.8, font.axis=2 (axis/tick text/labels)
* font.axis=2
* col.axis="grey50"
* 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 ===
=== layout ===
http://datascienceplus.com/adding-text-to-r-plot/
* [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 ===
=== reset the settings ===
<syntaxhighlight lang='rsplus'>
{{Pre}}
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)  
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
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
</syntaxhighlight>
</pre>


=== mtext (margin text) vs title ===
=== mtext (margin text) vs title ===
Line 5,542: Line 7,083:
* https://datascienceplus.com/mastering-r-plot-part-2-axis/
* https://datascienceplus.com/mastering-r-plot-part-2-axis/


=== mgp (axis label locations) ===
=== 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 ‘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.
# 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.
# 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).
#* 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 ===
=== pch and point shapes ===
[[:File:R pch.png]]
[[:File:R pch.png]]


Line 5,552: Line 7,105:


* Full circle: pch=16
* Full circle: pch=16
* Display all possibilities: ggpubr::show_point_shapes()


=== lty (line type) ===
=== lty (line type) ===
[[:File:R lty.png]]
[[: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].
See [http://www.sthda.com/english/wiki/line-types-in-r-lty here].
ggpubr::show_line_types()


=== las (label style) ===
=== las (label style) ===
Line 5,567: Line 7,125:
3: Always vertical
3: Always vertical


=== oma (outer margin), common title for two plots ===
=== oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots ===
The following trick is useful when we want to draw multiple plots with a common title.
<ul>
 
<li>The following trick is useful when we want to draw multiple plots with a common title.
<syntaxhighlight lang='rsplus'>
{{Pre}}
par(mfrow=c(1,2),oma = c(0, 0, 2, 0))  # oma=c(0, 0, 0, 0) by default
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(1:10,  main="Plot 1")
plot(1:100,  main="Plot 2")
plot(1:100,  main="Plot 2")
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
</syntaxhighlight>
</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>


[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
=== 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 ==
== Non-standard fonts in postscript and pdf graphics ==
Line 5,586: Line 7,158:
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/


== save() vs saveRDS() ==
== 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() 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].
# 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].
Line 5,596: Line 7,168:
</pre>
</pre>


== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
[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''').
* ==: exact match
{{Pre}}
* all.equal: compare R objects x and y testing ‘near equality’
$ data(pbc, package = "survival")
* identical: The safe and reliable way to test two objects for being exactly equal.
$ names(pbc)
<syntaxhighlight lang='rsplus'>
$ 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
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
all.equal(x, y)
Line 5,606: Line 7,200:
identical(x, y)
identical(x, y)
# [1] FALSE
# [1] FALSE
</syntaxhighlight>
</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.
See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.


== testhat ==
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
* https://github.com/r-lib/testthat
* [http://www.win-vector.com/blog/2019/03/unit-tests-in-r/ Unit Tests in R]
* [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 ==
== tinytest ==
Line 5,621: Line 7,234:
== Numerical Pitfall ==
== Numerical Pitfall ==
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
<syntaxhighlight lang='bash'>
{{Pre}}
.1 - .3/3
.1 - .3/3
## [1] 0.00000000000000001388
## [1] 0.00000000000000001388
</syntaxhighlight>
</pre>


== Sys.getpid() ==
== 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].
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 ==
== How to write R codes ==
Line 5,635: Line 7,258:
** return and exit early,
** return and exit early,
** polymorphism,
** polymorphism,
** switch(),
** 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(),
** case_when(),
** %||%.
** %||%.
 
* [https://appsilon.com/write-clean-r-code/ 5 Tips for Writing Clean R Code] – Leave Your Code Reviewer Commentless
== dput: Writes an ASCII text representation of an R object ==
** Comments
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/dput dput]
** Strings
<pre>
** Loops
$ data(pbc, package = "survival")
** Code Sharing
$ names(pbc)
**Good Programming Practices
$ 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")
</pre>


== How to debug an R code ==
== How to debug an R code ==
Line 5,671: Line 7,289:


On Mac & R 3.4.0 (it's fine)
On Mac & R 3.4.0 (it's fine)
<syntaxhighlight lang='rsplus'>
{{Pre}}
> Sys.getenv("R_LIBS_USER")
> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
[1] "~/Library/R/3.4/library"
> .libPaths()
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
</syntaxhighlight>
</pre>


On Linux & R 3.3.1 (ARM)
On Linux & R 3.3.1 (ARM)
<syntaxhighlight lang='rsplus'>
{{Pre}}
> Sys.getenv("R_LIBS_USER")
> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
Line 5,685: Line 7,303:
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"
[2] "/usr/local/lib/R/library"
</syntaxhighlight>
</pre>


On Linux & R 3.4.1 (*Problematic*)
On Linux & R 3.4.1 (*Problematic*)
<syntaxhighlight lang='rsplus'>
{{Pre}}
> Sys.getenv("R_LIBS_USER")
> Sys.getenv("R_LIBS_USER")
[1] ""
[1] ""
Line 5,694: Line 7,312:
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"
[3] "/usr/lib/R/library"
</syntaxhighlight>
</pre>


I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
<syntaxhighlight lang='rsplus'>
{{Pre}}
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
> library(devtools)
Line 5,712: Line 7,330:
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works
> library(devtools) # Works
</syntaxhighlight>
</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].
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].
Line 5,725: Line 7,343:
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script 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.'' '''
* [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 ==
== Decimal point & decimal comma ==
Line 5,738: Line 7,357:
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]


== Random numbers: multivariate normal ==
== Random number generator ==
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
* 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  
The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See  
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
* [https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
<ul>
<syntaxhighlight lang='rsplus'>
<li>[https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
set.seed(1234)
 
junk <- biospear::simdata(n=500, p=500, q.main = 10, q.inter = 10,
See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754 this example]. A little more investigation shows the eigen values differ a little bit on macOS and Linux. See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754#file-mvtnorm_debug-r here].
                          prob.tt = .5, m0=1, alpha.tt= -.5,
</li>
                          beta.main= -.5, beta.inter= -.5, b.corr = .7, b.corr.by=25,
</ul>
                          wei.shape = 1, recr=3, fu=2, timefactor=1)
 
## Method 1: MASS::mvrnorm()
== rle() running length encoding ==
## This is simdata() has used. It gives different numbers on different OS.
* https://en.wikipedia.org/wiki/Run-length_encoding
##
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
library(MASS)
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
set.seed(1234)
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
n <- 500
* https://rosettacode.org/wiki/Run-length_encoding
p <- 500
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
b.corr.by <- 25
* 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
b.corr <- .7
 
n.blocks <- p%/%b.corr.by
== citation() ==
covMat <- diag(n.blocks) %x%
{{Pre}}
  matrix(b.corr^abs(matrix(1:b.corr.by, b.corr.by, b.corr.by, byrow = TRUE) -  
citation()
                    matrix(1:b.corr.by, b.corr.by, b.corr.by)), b.corr.by, b.corr.by)
citation("MASS")
diag(covMat) <- 1
toBibtex(citation())
x <- MASS::mvrnorm(n, rep(0, p), Sigma = covMat)
</pre>
range(x)
 
eS <- eigen(covMat, symmetric = TRUE) # used by MASS::mvrnorm()
== R not responding request to interrupt stop process ==
eS$vectors[which(eS$vectors[, 500] != 0), 500] # mac and Linux have opposite sign
[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).
 
 
packageVersion("MASS")
== Monitor memory usage ==
# Mac: [1] ‘7.3.49’
* x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
# Linux: [1] ‘7.3.49’
* Windows: memory.size(max=TRUE)
# Windows: [1] ‘7.3.47’
* 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.version$version.string
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
# Mac: [1] "R version 3.4.3 (2017-11-30)"
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
# Linux: [1] "R version 3.4.4 (2018-03-15)"
* '''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().
# Windows: [1] "R version 3.4.3 (2017-11-30)"
* [https://cran.r-project.org/web/packages/peakRAM/index.html peakRAM]: Monitor the Total and Peak RAM Used by an Expression or Function
 
* [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]
## Method 2: mvtnorm::rmvnorm()
 
set.seed(1234)
References:
x <- mvtnorm::rmvnorm(n=n, rep(0, p), sigma=covMat)
* [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 $PID'' is recommended. It only shows the percentage of memory usage.
# NOTE: eigen is used by default
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
range(x)
* [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).
# Mac: [1] -4.482566  4.459236
* [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 '''
# Linux: [1] -4.482566  4.459236
 
 
== Monitor Data ==
## Method 3: mvnfast::rmvn()
[https://www.jstatsoft.org/article/view/v098i01?s=09 Monitoring Data in R with the lumberjack Package]
set.seed(1234)
 
x <- mvnfast::rmvn(n, rep(0, p), covMat)
== Pushover ==
range(x)
[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}]
# Mac: [1] -4.323585  4.355666
 
# Linux: [1] -4.323585  4.355666
[https://cran.r-project.org/web/packages/pushoverr/ pushoverr]
 
 
library(microbenchmark)
= Resource =
library(MASS)
== Books ==
library(mvtnorm)
* [https://forwards.github.io/rdevguide/ R Development Guide] R Contribution Working Group
library(mvnfast)
* [https://rviews.rstudio.com/2021/11/04/bookdown-org/ An R Community Public Library] 2011-11-04
microbenchmark(v1 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "eigen"),
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
              v2 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "svd"),
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
              v3 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "chol"),
* [http://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
              v4 <- rmvn(n, rep(0, p), covMat),
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
              v5 <- mvrnorm(n, rep(0, p), Sigma = covMat))
* [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
Unit: milliseconds
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
expr      min        lq
* [https://b-rodrigues.github.io/fput/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
v1 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "eigen") 296.55374 300.81089
* [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
v2 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "svd") 461.81867 466.98806
* [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
v3 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "chol") 118.33759 120.01829
* [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]
v4 <- rmvn(n, rep(0, p), covMat)  66.64675  69.89383
* [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
v5 <- mvrnorm(n, rep(0, p), Sigma = covMat) 291.19826 294.88038
* [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
mean    median        uq      max neval  cld
* [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
306.72485 301.99339 304.46662 335.6137  100    d
* [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
478.58536 470.44085 493.89041 571.7990  100    e
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
125.85427 121.26185 122.21361 151.1658  100  b 
* Data Manipulation with R by Phil Spector
71.67996  70.52985  70.92923 100.2622  100 a  
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
301.88144 296.76028 299.50839 346.7049  100  c 
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
</syntaxhighlight>
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
A little more investigation shows the eigen values differ a little bit on macOS and Linux.
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
<syntaxhighlight lang='rsplus'>
<ul>
set.seed(1234); x <- mvrnorm(n, rep(0, p), Sigma = covMat)
<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.
debug(mvrnorm)  
{{Pre}}
# eS --- macOS
# R 3.4.1
# eS2 -- Linux
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
Browse[2]> range(abs(eS$values - eS2$values))
setwd("/tmp/efficientR/")
# [1] 0.000000e+00 1.776357e-15
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
Browse[2]> var(as.vector(eS$vectors))
# generated pdf file is located _book/_main.pdf
[1] 0.002000006
 
Browse[2]> var(as.vector(eS2$vectors))
bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
[1] 0.001999987
# generated epub file is located _book/_main.epub.
Browse[2]> all.equal(eS$values, eS2$values)
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
[1] TRUE
# but it is OK to run in an R terminal
Browse[2]> which(eS$values != eS2$values)
</pre>
  [1]  6  7  8  9  10  11  12  13  14  20  22  23  24  25  26  27  28  29
</li>
  ...
</ul>
[451] 494 495 496 497 499 500
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro
Browse[2]> range(abs(eS$vectors - eS2$vectors))
* [https://rstats.wtf/ What They Forgot to Teach You About R] Jennifer Bryan & Jim Hester
[1] 0.0000000 0.5636919
* [http://knosof.co.uk/ESEUR/ Evidence-based Software Engineering] by Derek M. Jones
</syntaxhighlight>
* [https://www.bigbookofr.com/index.html Big Book of R]
 
* [https://epirhandbook.com/?s=09 R for applied epidemiology and public health]
== rle() running length encoding ==
* [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://en.wikipedia.org/wiki/Run-length_encoding
* [https://education.rstudio.com/learn/ RStudio &rarr; Finding Your Way To R]. Beginners/Intermediates/Experts
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
* [https://deepr.gagolewski.com/index.html Deep R Programming]
* [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]
== Videos ==
* https://rosettacode.org/wiki/Run-length_encoding
[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.
* 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
=== Webinar ===
 
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
== citation() ==
 
<syntaxhighlight lang='rsplus'>
== useR! ==
citation()
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
citation("MASS")
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]
toBibtex(citation())
* [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]
</syntaxhighlight>
* [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]


== Monitor memory usage ==
[https://moderndata.plot.ly/why-dash-uses-the-mit-license/ Why Dash uses the mit license (and not a copyleft gpl license)]
* Windows: memory.size(max=TRUE)
* Linux
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
 
References:
* [https://unix.stackexchange.com/questions/554/how-to-monitor-cpu-memory-usage-of-a-single-process How to monitor CPU/memory usage of a single process?]. ''htop -p $PID'' is recommended. It only shows the percentage of memory usage.
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
* [https://serverfault.com/a/264856 How can I see the memory usage of a Linux process?] ''pmap $PID | tail -n 1'' is recommended. It shows the memory usage in absolute value (eg 1722376K).
 
= Resource =
== Books ==
* 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://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
* [http://www.amazon.com/Advanced-Chapman-Hall-CRC-Series/dp/1466586966/ref=pd_sim_b_6?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB Advanced R] by Hadley Wickham 2014
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
* [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/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
* [https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
: <syntaxhighlight lang='rsplus'>
# R 3.4.1
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
setwd("/tmp/efficientR/")
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# generated pdf file is located _book/_main.pdf


bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
== 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.