R: Difference between revisions

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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


See also more examples under [[C#Socket_Programming_Examples_using_C.2FC.2B.2B.2FQt|C page]].
As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See  some discussions in
 
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
=== Others ===
* http://www.mail-archive.com/rcpp-devel@lists.r-forge.r-project.org/msg04377.html
* http://rosettacode.org/wiki/Hello_world/ (Different languages)
So the Qt and Wt web tool applications on Windows may or may not be possible.
* http://kperisetla.blogspot.com/2012/07/simple-http-web-server-in-c.html (Windows web server)
* http://css.dzone.com/articles/web-server-c (handling HTTP GET request, handling content types(txt, html, jpg, zip. rar, pdf, php etc.), sending proper HTTP error codes, serving the files from a web root, change in web root in a config file, zero copy optimization using sendfile method and php file handling.)
* https://github.com/gtungatkar/Simple-HTTP-server
* https://github.com/davidmoreno/onion
 
== shiny ==
See [[Shiny|Shiny]].
 
== [https://www.rplumber.io/ plumber]: Turning your R code into a RESTful Web API ==
* https://github.com/trestletech/plumber
* 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 ==
== GUI ==
* There are two major Docker images. They include gcc, gfortran, .... So it can be used to install Rcpp package for example.
=== Qt and R ===
** [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.
* 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]
** [https://hub.docker.com/r/rocker/r-base rocker project] which only has the latest tag
* http://qtinterfaces.r-forge.r-project.org
* [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] ==
== tkrplot ==
http and WebSocket library.
On Ubuntu, we need to install tk packages, such as by
<pre>
sudo apt-get install tk-dev
</pre>


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.
== reticulate - Interface to 'Python' ==
[[Python#R_and_Python:_reticulate_package|Python -> reticulate]]


== [http://rapache.net/ RApache] ==
== Hadoop (eg ~100 terabytes) ==
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]


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


* http://www.jstatsoft.org/v49/i10/paper
=== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ===
* [https://github.com/jverzani/gWidgetsWWW2 gWidgetsWWW2] gWidgetsWWW based on Rook
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
* [http://www.r-statistics.com/2012/11/comparing-shiny-with-gwidgetswww2-rapache/ Compare shiny with gWidgetsWWW2.rapache]
* Ubuntu 12.04 - [http://crishantha.com/wp/?p=1414 Crishantha.com], [http://nikhilshah123sh.blogspot.com/2014/03/setting-up-rhadoop-in-ubuntu-1204.html nikhilshah123sh.blogspot.com].[http://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/ Bighadoop.wordpress] contains an example.
* RapReduce in R by [https://github.com/RevolutionAnalytics/rmr2/blob/master/docs/tutorial.md RevolutionAnalytics] with a few examples.
* https://twitter.com/hashtag/rhadoop
* [http://bigd8ta.com/step-by-step-guide-to-setting-up-an-r-hadoop-system/ Bigd8ta.com] based on Ubuntu 14.04.


== [http://cran.r-project.org/web/packages/Rook/index.html Rook] ==
=== 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


Since R 2.13, the internal web server was exposed.
== [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.
[https://docs.google.com/present/view?id=0AUTe_sntp1JtZGdnbjVicTlfMzFuZDQ5dmJxNw Tutorual from useR2012] and [https://github.com/rstats/RookTutorial Jeffrey Horner]
<pre>
 
sudo apt-get update
Here is another [http://www.rinfinance.com/agenda/2011/JeffHorner.pdf one] from http://www.rinfinance.com.
sudo apt-get install libxml2-dev
 
</pre>
Rook is also supported by [rApache too. See http://rapache.net/manual.html.
 
Google group. https://groups.google.com/forum/?fromgroups#!forum/rrook
 
Advantage
* the web applications are created on desktop, whether it is Windows, Mac or Linux.
* No Apache is needed.
* create [http://jeffreyhorner.tumblr.com/post/4723187316/introducing-rook multiple applications] at the same time. This complements the limit of rApache.  
 
----
 
4 lines of code [http://jeffreybreen.wordpress.com/2011/04/25/4-lines-of-r-to-get-you-started-using-the-rook-web-server-interface/ example].


On CentOS,
<pre>
<pre>
library(Rook)
yum -y install libxml2 libxml2-devel
s <- Rhttpd$new()
s$start(quiet=TRUE)
s$print()
s$browse(1)  # OR s$browse("RookTest")
</pre>
</pre>
Notice that after s$browse() command, the cursor will return to R because the command just a shortcut to open the web page http://127.0.0.1:10215/custom/RookTest.


[[File:Rook.png|100px]]
=== XML ===
[[File:Rook2.png|100px]]
* 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()'''.
[[File:Rookapprnorm.png|100px]]
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
* https://www.tutorialspoint.com/r/r_xml_files.htm
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
* [http://www.stat.berkeley.edu/~statcur/Workshop2/Presentations/XML.pdf Extracting data from XML] PubMed and Zillow are used to illustrate. xmlTreeParse(),  xmlRoot(),  xmlName() and xmlSApply().
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
{{Pre}}
library(XML)


We can add Rook '''application''' to the server; see ?Rhttpd.
# Read and parse HTML file
<pre>
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)
s$add(
    app=system.file('exampleApps/helloworld.R',package='Rook'),name='hello'
)
s$add(
    app=system.file('exampleApps/helloworldref.R',package='Rook'),name='helloref'
)
s$add(
    app=system.file('exampleApps/summary.R',package='Rook'),name='summary'
)


s$print()
# 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))


#Server started on 127.0.0.1:10221
# Replace all by spaces
#[1] RookTest http://127.0.0.1:10221/custom/RookTest
doc.text = gsub('\n', ' ', doc.text)
#[2] helloref http://127.0.0.1:10221/custom/helloref
#[3] summary  http://127.0.0.1:10221/custom/summary
#[4] hello    http://127.0.0.1:10221/custom/hello


# Stops the server but doesn't uninstall the app
# Join all the elements of the character vector into a single
## Not run:
# character string, separated by spaces
s$stop()
doc.text = paste(doc.text, collapse = ' ')
 
## End(Not run)
s$remove(all=TRUE)
rm(s)
</pre>
</pre>
For example, the interface and the source code of ''summary'' app are given below


[[File:Rookappsummary.png|100px]]
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.
 
{{Pre}}
<nowiki>
> library(RCurl) # getURL()
app <- function(env) {
> library(XML)   # htmlParse and xpathSApply
    req <- Rook::Request$new(env)
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
    res <- Rook::Response$new()
> doc = htmlParse(xData)
    res$write('Choose a CSV file:\n')
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
    res$write('<form method="POST" enctype="multipart/form-data">\n')
  # I look at the source code and search 2.5.3a and find the tag as
    res$write('<input type="file" name="data">\n')
  # <span class="css-truncate-target">2.5.3a</span>
    res$write('<input type="submit" name="Upload">\n</form>\n<br>')
> plain.text
 
[1] "2.5.3a"     "2.5.2b"     "2.5.2a"     "2.5.1b"     "2.5.1a"     
    if (!is.null(req$POST())){
[6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
data <- req$POST()[['data']]
>
res$write("<h3>Summary of Data</h3>");
> # try bwa
res$write("<pre>")
> > xData <- getURL("https://github.com/lh3/bwa/releases")
res$write(paste(capture.output(summary(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
> doc = htmlParse(xData)
res$write("</pre>")
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
res$write("<h3>First few lines (head())</h3>");
[1] "v0.7.15" "v0.7.13"
res$write("<pre>")
res$write(paste(capture.output(head(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
res$write("</pre>")
     }
    res$finish()
}
</nowiki>
 
More example:
* http://lamages.blogspot.com/2012/08/rook-rocks-example-with-googlevis.html
* [http://www.road2stat.com/cn/r/rook.html Self-organizing map]
* Deploy Rook apps with rApache. [http://jeffreyhorner.tumblr.com/post/27861973339/deploy-rook-apps-with-rapache-part-i First one] and [http://jeffreyhorner.tumblr.com/post/33814488298/deploy-rook-apps-part-ii two].
* [https://rud.is/b/2016/07/05/a-simple-prediction-web-service-using-the-new-firery-package/ A Simple Prediction Web Service Using the New fiery Package]
 
== [https://code.google.com/p/sumo/ sumo] ==
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.


== [http://www.stat.ucla.edu/~jeroen/stockplot Stockplot] ==
> # 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"
</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").


== [http://www.rforge.net/FastRWeb/ FastRWeb] ==
=== xmlview ===
http://cran.r-project.org/web/packages/FastRWeb/index.html
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/


== WebDriver ==
== RCurl ==
'WebDriver' Client for 'PhantomJS'
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>


https://github.com/rstudio/webdriver
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


== [http://sysbio.mrc-bsu.cam.ac.uk/Rwui/tutorial/Instructions.html Rwui] ==
No google ID is required


== [http://cran.r-project.org/web/packages/CGIwithR/index.html CGHWithR] and [http://cran.r-project.org/web/packages/WebDevelopR/ WebDevelopR] ==
Seems not work
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>
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :  
  arguments imply differing number of rows: 2, 0
</pre>


== [http://www.rstudio.com/ide/docs/advanced/manipulate manipulate] from RStudio ==
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
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].
'''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


Mathematica also has manipulate function for plotting; see [http://reference.wolfram.com/mathematica/tutorial/IntroductionToManipulate.html here].
# Ubuntu 16.04, 18.04
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev


== [https://github.com/att/rcloud RCloud] ==
# Ubuntu 20.04
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.
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev
</pre>


See also the [http://user2014.stat.ucla.edu/abstracts/talks/193_Harner.pdf Talk] in UseR 2014.
[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/cloudyr cloudyr] and [https://github.com/socialcopsdev/flyio flyio] - Input Output Files in R from Cloud or Local ==
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.
https://blog.socialcops.com/inside-sc/announcements/flyio-r-package-interact-data-cloud/ Announcing flyio, an R Package to Interact with Data in the Cloud]


== Dropbox access ==
=== [https://github.com/hadley/httr httr] ===
[https://cran.r-project.org/web/packages/rdrop2/index.html rdrop2] package
httr imports curl, jsonlite, mime, openssl and R6 packages.


== Web page scraping ==
When I tried to install httr package, I got an error and some message:
http://www.slideshare.net/schamber/web-data-from-r#btnNext
<pre>
 
Configuration failed because openssl was not found. Try installing:
=== [https://cran.r-project.org/web/packages/xml2/ xml2] package ===
* deb: libssl-dev (Debian, Ubuntu, etc)
rvest package depends on xml2.
* 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’
</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!


=== [https://cran.r-project.org/web/packages/purrr/index.html purrr]: Functional Programming Tools ===
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://purrr.tidyverse.org/
* [http://colinfay.me/purrr-cookbook/ purrr cookbook]
* 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] ===
Since httr package was used in many other packages, take a look at how others use it. For example, [https://github.com/ropensci/aRxiv aRxiv] package.
[http://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!
[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)
<syntaxhighlight lang='bash'>
sudo apt-get install libcurl4-openssl-dev # OR libcurl4-gnutls-dev


sudo apt-get install libxml2-dev
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
</syntaxhighlight>
curl is independent of RCurl package.


* https://github.com/hadley/rvest
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
* [http://datascienceplus.com/visualizing-obesity-across-united-states-by-using-data-from-wikipedia/ Visualizing obesity across United States by using data from Wikipedia]
* https://www.opencpu.org/posts/curl-release-0-8/
* [https://stat4701.github.io/edav/2015/04/02/rvest_tutorial/ rvest tutorial: scraping the web using R]
* https://renkun.me/pipeR-tutorial/Examples/rvest.html
* http://zevross.com/blog/2015/05/19/scrape-website-data-with-the-new-r-package-rvest/
* [https://datascienceplus.com/google-scholar-scraping-with-rvest/ Google scholar scraping with rvest package]


=== Animate ===
{{Pre}}
* [https://guyabel.com/post/football-kits/ Animating Changes in Football Kits using R]: rvest, tidyverse, xml2, purrr & magick
library(curl)
* [https://guyabel.com/post/animated-directional-chord-diagrams/ Animated Directional Chord Diagrams] tweenr & magick
h <- new_handle()
* [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]
handle_setform(h,
* [https://www.listendata.com/2019/05/create-animation-in-r-learn-with.html Create animation in R]: learn by examples (gganimate)
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)
</pre>


=== [https://cran.r-project.org/web/packages/V8/index.html V8]: Embedded JavaScript Engine for R ===
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
[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.
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment


=== [http://cran.r-project.org/web/packages/pubmed.mineR/index.html pubmed.mineR] ===
== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
Text mining of PubMed Abstracts (http://www.ncbi.nlm.nih.gov/pubmed). The algorithms are designed for two formats (text and XML) from PubMed.
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).


[https://github.com/jtleek/swfdr R code for scraping the P-values from pubmed, calculating the Science-wise False Discovery Rate, et al] (Jeff Leek)
Example:
{{Pre}}
# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')
</pre>


== These R packages import sports, weather, stock data and more ==
== DirichletMultinomial ==
* https://www.computerworld.com/article/3109890/data-analytics/these-r-packages-import-sports-weather-stock-data-and-more.html
On Ubuntu, we do
* https://github.com/ALShum/rwunderground
<pre>
* [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.
sudo apt-get install libgsl0-dev
* https://cran.r-project.org/web/packages/rnoaa/index.html. A personal API key (token) is required. 10,000 requests per day
</pre>
* <strike>http://ram-n.github.io/weatherData/ </strike> (not working)


== Diving Into Dynamic Website Content with splashr ==
== Create GUI ==
https://rud.is/b/2017/02/09/diving-into-dynamic-website-content-with-splashr/
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===


== Send email ==
== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
=== [https://github.com/rpremraj/mailR/ mailR] ===
[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]
Easiest. Require rJava package (not trivial to install, see [[#RJava|rJava]]). mailR is an interface to Apache Commons Email to send emails from within R. See also [http://unamatematicaseltigre.blogspot.com/2016/12/how-to-send-bulk-email-to-your-students.html send bulk email]


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


[https://r-bar.net/mailr-smtp-webmail-starttls-tls-ssl/ MailR SMTP Setup (Gmail, Outlook, Yahoo) | STARTTLS]
== Map ==
=== [https://rstudio.github.io/leaflet/ leaflet] ===
* rstudio.github.io/leaflet/#installation-and-use
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


=== [https://cran.r-project.org/web/packages/gmailr/index.html gmailr] ===
=== choroplethr ===
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://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
<syntaxhighlight lang='rsplus'>
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
library(gmailr)
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/
gmail_auth('mysecret.json', scope = 'compose')


test_email <- mime() %>%
=== ggplot2 ===
  to("to@gmail.com") %>%
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]
  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] ===  
== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
sendmailR provides a simple SMTP client. It is not clear how to use the package (i.e. where to enter the password).
See an example from [[R#RJSONIO|RJSONIO]] above.


=== emayili ===
== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
[https://datawookie.netlify.com/blog/2019/05/emayili-sending-email-from-r/ emayili: Sending Email from R]
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.


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


== Interactive html output ==
== quantmod ==
=== [http://cran.r-project.org/web/packages/sendplot/index.html sendplot] ===
[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/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].
# Initial data downloading
# Update existing data
# Create a batch file


=== r2d3 ===
== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
[https://blog.rstudio.com/2018/10/05/r2d3-r-interface-to-d3-visualizations/ r2d3] - R Interface to D3 Visualizations
* 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/d3Network/index.html d3Network] ===
== Tool for connecting Excel with R ==
* http://christophergandrud.github.io/d3Network/ (old)
* https://bert-toolkit.com/
* https://christophergandrud.github.io/networkD3/ (new)
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
<source lang="rsplus">
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html
library(d3Network)


Source <- c("A", "A", "A", "A", "B", "B", "C", "C", "D")  
== write.table ==
Target <- c("B", "C", "D", "J", "E", "F", "G", "H", "I")  
=== Output a named vector ===
NetworkData <- data.frame(Source, Target)  
<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)


d3SimpleNetwork(NetworkData, height = 800, width = 1024, file="tmp.html")
# one liner: row names of a 'matrix' become the names of a vector
</source>
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
all.equal(vec, vec3)
</pre>


=== [http://cran.r-project.org/web/packages/htmlwidgets/ htmlwidgets for R] ===
=== Avoid leading empty column to header ===
Embed widgets in R Markdown documents and Shiny web applications.
[https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
<pre>
write.table(a, 'a.txt', col.names=NA)
</pre>


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


=== [http://cran.r-project.org/web/packages/networkD3/index.html networkD3] ===
=== read.delim(, row.names=1) and write.table(, row.names=TRUE) ===
This is a port of Christopher Gandrud's [http://christophergandrud.github.io/d3Network/ d3Network] package to the htmlwidgets framework.
[https://www.statology.org/read-delim-in-r/ How to Use read.delim Function in R]


=== [http://cran.r-project.org/web/packages/scatterD3/index.html scatterD3] ===
Case 1: no row.names
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.
<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>


=== dygraphs ===
== Read/Write Excel files package ==
* https://github.com/rstudio/dygraphs
* http://www.milanor.net/blog/?p=779
* [https://www.r-bloggers.com/visualizations-for-algorithmic-trading-in-r/ Visualizations for Algorithmic Trading in R]
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
 
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
=== [https://github.com/bwlewis/rthreejs rthreejs] - Create interactive 3D scatter plots, network plots, and globes ===
** [https://stackoverflow.com/a/17976604 Export both Image and Data from R to an Excel spreadsheet]
[http://bwlewis.github.io/rthreejs/ Examples]
* [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.
=== d3heatmap ===
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
See [[Heatmap#d3heatmap|R]]
** [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. 
=== [https://cran.r-project.org/web/packages/svgPanZoom/index.html svgPanZoom] ===
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
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.
** [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.'''
=== DT: An R interface to the DataTables library ===
** For writing to Excel formats, use writexl or openxlsx package.
* http://blog.rstudio.org/2015/06/24/dt-an-r-interface-to-the-datatables-library/
:<syntaxhighlight lang='rsplus'>
 
library(readxl)
=== plotly ===
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE,
* [http://moderndata.plot.ly/power-curves-r-plotly-ggplot2/ Power curves] and ggplot2.
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf,
* [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]
    guess_max = min(1000, n_max), progress = readxl_progress(),
* [http://moderndata.plot.ly/filled-chord-diagram-in-r-using-plotly/ Filled chord diagram]
    .name_repair = "unique")
* [https://moderndata.plot.ly/dashboards-in-r-with-shiny-plotly/ DASHBOARDS IN R WITH SHINY & PLOTLY]
# Example
* [https://plot.ly/r/shiny-tutorial/ Plotly Graphs in Shiny],
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
** [https://plot.ly/r/shiny-gallery/ Gallery]
** [https://plot.ly/r/shinyapp-UN-simple/ Single time series]
** [https://plot.ly/r/shinyapp-UN-advanced/ Multiple time series]
* [https://www.r-exercises.com/2017/09/28/how-to-plot-basic-charts-with-plotly/ How to plot basic charts with plotly]
* [https://www.displayr.com/how-to-add-trend-lines-in-r-using-plotly/?utm_medium=Feed&utm_source=Syndication How to add Trend Lines in R Using Plotly]
* [https://blog.methodsconsultants.com/posts/introduction-to-interactive-graphics-in-r-with-plotly/ Introduction to Interactive Graphics in R with plotly]
 
== Amazon ==
[https://github.com/56north/Rmazon Download product information and reviews from Amazon.com]
<syntaxhighlight lang='bash'>
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
</syntaxhighlight>
and in R
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R
<syntaxhighlight lang='rsplus'>
:<syntaxhighlight lang='rsplus'>
install.packages("devtools")
library(writexl)
install.packages("XML")
mylst <- list(sheet1name = df1, sheet2name = df2)
install.packages("pbapply")
write_xlsx(mylst, "output.xlsx")
install.packages("dplyr")
devtools::install_github("56north/Rmazon")
product_info <- Rmazon::get_product_info("1593273843")
reviews <- Rmazon::get_reviews("1593273843")
reviews[1,6] # only show partial characters from the 1st review
nchar(reviews[1,6])
as.character(reviews[1,6]) # show the complete text from the 1st review
 
reviews <- Rmazon::get_reviews("B07BNGJXGS")
# Fetching 30 reviews of 'BOOX Note Ereader,Android 6.0 32 GB 10.3" Dual Touch HD Display'
#  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed = 02s
reviews
# A tibble: 30 x 6
  reviewRating reviewDate reviewFormat Verified_Purcha… reviewHeadline
          <dbl> <chr>      <lgl>        <lgl>            <chr>       
1            4 May 23, 2… NA          TRUE            Good for PDF …
2            3 May 8, 20… NA          FALSE            The reading s…
3            5 May 17, 2… NA          TRUE            E-reader and …
4            3 May 24, 2… NA          TRUE            Good hardware…
5            3 June 21, … NA          TRUE            Poor QC     
6            5 August 5,… NA          TRUE            Excellent for…
7            5 May 31, 2… NA          TRUE            Especially li…
8            5 July 4, 2… NA          TRUE            Android 6 rea…
9            4 July 15, … NA          TRUE            Remember the …
10            4 June 9, 2… NA          TRUE            Overall fanta…
# ... with 20 more rows, and 1 more variable: reviewText <chr>
reviews[1, 6] # 6-th column is the review text
</syntaxhighlight>
</syntaxhighlight>


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


= Creating local repository for CRAN and Bioconductor =
The hidden worksheets become visible (Not sure what are those first rows mean in the output).
[[R_repository|R repository]]
{{Pre}}
> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00
[1] "Experiment descriptors" "Filtered log ratio"    "Gene identifiers"     
[4] "Gene annotations"      "CollateInfo"            "GeneSubsets"         
[7] "GeneSubsetsTemp"     
</pre>
 
The Chinese character works too.
{{Pre}}
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</pre>


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


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


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


== Bioconductor on Amazon EC2 ==
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.
http://www.bioconductor.org/help/bioconductor-cloud-ami/
* 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!)


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


== bigmemory, biganalytics, bigtabulate ==
Note that '''data.table::fread()''' can read a selection of the columns.


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


== data.table ==
== Data Manipulation & Tidyverse ==
See [[#data.table_2|data.table]].
See [[Tidyverse|Tidyverse]].


= Reproducible Research =
== Data Science ==
http://cran.r-project.org/web/views/ReproducibleResearch.html
See [[Data_science|Data science]] page


== Reproducible Environments ==
== microbenchmark & rbenchmark ==
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
* [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)


== Some lessons in R coding ==
== Plot, image ==
# don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use [[Rcpp|Rcpp]] package and R's random number generator instead.
=== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ===
# 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!
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.


= Useful R packages =
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].
* [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]


== Rcpp ==
=== png and resolution ===
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].
It seems people use '''res=300''' as a definition of high resolution.  


== RInside : embed R in C++ code ==
<ul>
* http://dirk.eddelbuettel.com/code/rinside.html
<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.
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
<li>[https://stackoverflow.com/a/51194014 Saving high resolution plot in png].
<pre>
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.


=== Ubuntu ===
# It seems the following command gives the same result as above
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.  
png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
</pre>
# Install [[Qt|Qt]]. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
<li>Chapter 14.5 [https://r-graphics.org/recipe-output-bitmap Outputting to Bitmap (PNG/TIFF) Files] by R Graphics Cookbook
# 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.  
* Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.  
# 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.
<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
[[File:qtdensity.png|100px]].
* 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>


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
=== 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>
<pre>
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
svg("svg4.svg", width=4, height=4)
make
plot(1:10, main="width=4, height=4")
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
dev.off()
 
svg("svg7.svg", width=7, height=7) # default
plot(1:10, main="width=7, height=7")
dev.off()
</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]).
</ul>
 
=== magick ===
https://cran.r-project.org/web/packages/magick/
 
See an example [[:File:Progpreg.png|here]] I created.
 
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
 
=== geDevices ===
* [https://www.jumpingrivers.com/blog/r-graphics-cairo-png-pdf-saving/ Saving R Graphics across OSs]. Use png(type="cairo-png") or the [https://cran.r-project.org/web/packages/ragg/index.html ragg] package which can be incorporated into RStudio.
* [https://www.jumpingrivers.com/blog/r-knitr-markdown-png-pdf-graphics/ Setting the Graphics Device in a RMarkdown Document]
 
=== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ===
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).


=== Windows 7 ===
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
To make RInside works on Windows OS, try the following
# Make sure R is installed under '''C:\''' instead of '''C:\Program Files''' if we don't want to get an error like ''g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory''.
# 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>
<pre>
@echo off
sudo apt-get install libgtk2.0-dev libcairo2-dev
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>


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


== GUI ==
=== sketcher: photo to sketch effects ===
=== Qt and R ===
https://htsuda.net/sketcher/
* 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 ==
=== httpgd ===
On Ubuntu, we need to install tk packages, such as by
* https://nx10.github.io/httpgd/ A graphics device for R that is accessible via network protocols. Display graphics on browsers.
<pre>
* [https://youtu.be/uxyhmhRVOfw Three tricks to make IDEs other than Rstudio better for R development]
sudo apt-get install tk-dev
</pre>


== reticulate - Interface to 'Python' ==
== [http://igraph.org/r/ igraph] ==
* https://cran.r-project.org/web/packages/reticulate/index.html, [https://github.com/rstudio/reticulate Github]
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]
** 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>
== Identifying dependencies of R functions and scripts ==
---
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
title: "R Notebook"
{{Pre}}
output: html_notebook
library(mvbutils)
---
foodweb(where = "package:batr")


```{r}
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
library(reticulate)
py_discover_config()
x <- 5
source_python("test.py")
y <- add_three(x)
print(y)
```


Pass R variables to Python. Works
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
```{python}
a = 7
print(r.x)
```
 
Pass python variables to R. Works.
```{r}
py$a
py_run_string("y = 10"); py$y
```
</pre>
</pre>


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


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


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


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


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


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


=== XML ===
=== [http://rpubs.com/gaston/colortools colortools] ===
* http://giventhedata.blogspot.com/2012/06/r-and-web-for-beginners-part-ii-xml-in.html. It gave an example of extracting the XML-values from each XML-tag for all nodes and save them in a data frame using '''xmlSApply()'''.
Tools that allow users generate color schemes and palettes
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
 
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
=== [https://github.com/daattali/colourpicker colourpicker] ===
* https://www.tutorialspoint.com/r/r_xml_files.htm
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
* 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().
=== eyedroppeR ===
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]
<syntaxhighlight lang='rsplus'>
library(XML)


# Read and parse HTML file
== [https://github.com/kevinushey/rex rex] ==
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)
Friendly Regular Expressions


# Extract all the paragraphs (HTML tag is p, starting at
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
# the root of the document). Unlist flattens the list to
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


# Replace all by spaces
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
doc.text = gsub('\n', ' ', doc.text)


# Join all the elements of the character vector into a single
<pre>
# character string, separated by spaces
library(formatR)
doc.text = paste(doc.text, collapse = ' ')
tidy_source("Input.R", file = "output.R", width.cutoff=70)
</syntaxhighlight>
tidy_source("clipboard")
# default width is getOption("width") which is 127 in my case.
</pre>


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.
Some issues
<syntaxhighlight lang='rsplus'>
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
> 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
> 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"
</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://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/
 
== RCurl ==
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
<syntaxhighlight lang='bash'>
# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
 
=== Scrape google scholar results ===
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R
 
No google ID is required
 
Seems not work
<pre>
<pre>
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']", :
cat("abcd",
  arguments imply differing number of rows: 2, 0
    # This is my comment
    "defg")
</pre>
</pre>
 
will result in
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
<pre>
'''devtools''' package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the [[#remotes|remotes]] package which was suggested by the [https://cran.r-project.org/web/packages/BiocManager/index.html BiocManager] package.
> tidy_source("clipboard")
<syntaxhighlight lang='bash'>
Error in base::parse(text = code, srcfile = NULL) :
# Ubuntu 14.04
  3:1: unexpected string constant
sudo apt-get install libcurl4-openssl-dev
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
 
3: "defg"
# Ubuntu 16.04
  ^
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev
</pre>
</syntaxhighlight>
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.  
 
<pre>
[https://github.com/wch/movies/issues/3 Lazy-load database XXX is corrupt. internal error -3]. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.
cat("abcd"
 
    ,"defg"  # This is my comment
=== [https://github.com/hadley/httr httr] ===
  ,"ghij")
httr imports curl, jsonlite, mime, openssl and R6 packages.
</pre>
 
will become
When I tried to install httr package, I got an error and some message:
<pre>
<pre>
Configuration failed because openssl was not found. Try installing:
cat("abcd", "defg"  # This is my comment
* deb: libssl-dev (Debian, Ubuntu, etc)
, "ghij")  
* rpm: openssl-devel (Fedora, CentOS, RHEL)
* 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!
Still bad!!
 
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
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).
<pre>
 
cat("</p>",
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.
"<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>
 
== styler ==
https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.


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


* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
* https://www.opencpu.org/posts/curl-release-0-8/
Interface to the arXiv API


<syntaxhighlight lang='rsplus'>
=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
library(curl)
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
h <- new_handle()
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
handle_setform(h,
* https://ropensci.org/technotes/2018/12/14/pdftools-20/
  name="aaa", email="bbb"
 
)
== [https://github.com/ColinFay/aside aside]: set it aside ==
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
An RStudio addin to run long R commands aside your current session.
rawToChar(req$content)
 
</syntaxhighlight>
== Teaching ==
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics


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


== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).
[https://rcrastinate.rbind.io/post/how-to-save-and-load-data-in-r-an-overview/ How to save (and load) datasets in R: An overview]


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


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


== Create GUI ==
=== Efficient Data Management in R ===
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.


== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
== Text to speech ==
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


== [http://cran.r-project.org/web/packages/rjson/index.html rjson] ==
== Speech to text ==
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/
https://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]


== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ==
== Weather data ==
=== Accessing Bitcoin Data with R ===
* [https://github.com/ropensci/prism prism] package
http://blog.revolutionanalytics.com/2015/11/accessing-bitcoin-data-with-r.html
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


=== Plot IP on google map ===
== logR ==
* http://thebiobucket.blogspot.com/2011/12/some-fun-with-googlevis-plotting-blog.html#more  (RCurl, RJONIO, plyr, googleVis)
https://github.com/jangorecki/logR
* 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.
== Progress bar ==
<pre>
https://github.com/r-lib/progress#readme
require(RJSONIO) # fromJSON
require(RCurl)  # getURL


temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
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'.
                          ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
names(ip) <- "IP"
nr = nrow(ip)
Lon <- as.numeric(rep(NA, nr))
Lat <- Lon
Coords <- data.frame(Lon, Lat)
ip2coordinates <- function(ip) {
  api <- "http://freegeoip.net/json/"
  get.ips <- getURL(paste(api, URLencode(ip), sep=""))
  # result <- ldply(fromJSON(get.ips), data.frame)
  result <- data.frame(fromJSON(get.ips))
  names(result)[1] <- "ip.address"
  return(result)
}


for (i in 1:nr){
== cron ==
  cat(i, "\n")
* [https://github.com/bnosac/cronr cronR]
  try(
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]
  Coords[i, 1:2] <- ip2coordinates(ip$IP[i])[c("longitude", "latitude")]
  )
}
# append to log-file:
logfile <- data.frame(ip, Lat = Coords$Lat, Long = Coords$Lon,
                                      LatLong = paste(round(Coords$Lat, 1), round(Coords$Lon, 1), sep = ":"))
log_gmap <- logfile[!is.na(logfile$Lat), ]


require(googleVis) # gvisMap
== beepr: Play A Short Sound ==
gmap <- gvisMap(log_gmap, "LatLong",
https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".
                options = list(showTip = TRUE, enableScrollWheel = TRUE,
                              mapType = 'hybrid', useMapTypeControl = TRUE,
                              width = 1024, height = 800))
plot(gmap)
</pre>
[[File:GoogleVis.png|200px]]


The plot.gvis() method in googleVis packages also teaches the startDynamicHelp() function in the tools package, which was used to launch a http server. See
== utils package ==
[http://jeffreyhorner.tumblr.com/page/3 Jeffrey Horner's note about deploying Rook App].
https://www.rdocumentation.org/packages/utils/versions/3.6.2


== Map ==
== tools package ==
=== [https://rstudio.github.io/leaflet/ leaflet] ===
* https://www.rdocumentation.org/packages/tools/versions/3.6.2
* rstudio.github.io/leaflet/#installation-and-use
* [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>)]
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


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


== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
== R call C/C++ ==
See an example from [[R#RJSONIO|RJSONIO]] above.
Mainly talks about .C() and .Call().


== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
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.
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.


== gtrendsR - Google Trends ==
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
* [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
* http://www.stat.harvard.edu/ccr2005/
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=dna%20sequencing dna sequencing] from 2004-01-01.
* http://mazamascience.com/WorkingWithData/?p=1099
* [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.
* [https://youtube.com/playlist?list=PLwc48KSH3D1OkObQ22NHbFwEzof2CguJJ Make an R package with C++ code] (a playlist from youtube)
* [http://www.kdnuggets.com/2017/09/python-vs-r-data-science-machine-learning.html?utm_content=buffere1df7&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer Python vs R – Who Is Really Ahead in Data Science, Machine Learning?] and [https://stackoverflow.blog/2017/09/06/incredible-growth-python/ The Incredible Growth of Python] by [https://twitter.com/drob?lang=en David Robinson]
* [https://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
 
=== .Call ===
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/CallExternal ?.Call]
* [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”)]


== quantmod ==
Be sure to add the ''PACKAGE'' parameter to avoid an error like
[http://www.thertrader.com/2015/12/13/maintaining-a-database-of-price-files-in-r/ Maintaining a database of price files in R]. It consists of 3 steps.
<pre>
cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group,
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) :  
  "standardize" not resolved from current namespace (grpreg)
</pre>


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


== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
(From [https://cran.r-project.org/doc/manuals/r-release/R-exts.html#dyn_002eload-and-dyn_002eunload Writing R Extensions manual]) Loading is most often done automatically based on the '''useDynLib()''' declaration in the '''NAMESPACE''' file, but may be done explicitly via a call to '''library.dynam()'''. This has the form
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
{{Pre}}
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
library.dynam("libname", package, lib.loc)
* https://github.com/cran/caret source code mirrored on github
</pre>
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/


== Tool for connecting Excel with R ==
=== library.dynam.unload() ===
* https://bert-toolkit.com/
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
* http://r-pkgs.had.co.nz/src.html. The '''library.dynam.unload()''' function should be placed in '''.onUnload()''' function. This function can be saved in any R files.
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]


== Read/Write Excel files package ==
=== gcc ===
* http://www.milanor.net/blog/?p=779
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
 
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
=== Primitive functions ===
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]
** It can't be used to open .xls or .xlm files.
 
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
== SEXP ==
** [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]
Some examples from packages
* [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'>
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
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>
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 Fortran ==
<syntaxhighlight lang='rsplus'>
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
> excel_sheets("~/Downloads/BRCA.xls")
* [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)
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.
== Embedding R ==
<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
* 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.
<syntaxhighlight lang='rsplus'>
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
read_excel_allsheets <- function(filename) {
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report] by Friedrich Leisch in 2007.
    sheets <- readxl::excel_sheets(filename)
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</syntaxhighlight>


== [https://cran.r-project.org/web/packages/readr/ readr] (it is not designed to read Excel files) ==
=== 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>.


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.
This example can be run by
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>


[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released.
Note:  
# '''R_HOME/bin/exec/R''' is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
# '''R_HOME/bin/R''' is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to ''/usr/local/bin/R''. When we run ''R_HOME/bin/R'', it actually runs ''R_HOME/bin/R CMD R_HOME/bin/exec/R'' (see line 259 of ''R_HOME/bin/R'' as in R 3.0.2) so we know the important role of ''R_HOME/bin/exec/R''.


The '''read_csv()''' function from the '''readr''' package is as fast as '''fread()''' function from '''data.table''' package. ''For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv().'' See 5.3 of Efficient R Programming book.
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.


Note that '''fread()''' can read-n a selection of the columns.
=== 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]].
== Colors ==
* 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].
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* [http://colorspace.r-forge.r-project.org/articles/colorspace.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes.
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c
* [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.  
Example:
Create [https://gist.github.com/arraytools/7d32d92fee88ffc029365d178bc09e75#file-embed-c embed.c] file.
Then build the executable. Note that I don't need to create R_HOME variable.
<pre>
cd
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest


According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
nano embed.c
# Using a single line will give an error and cannot not show the real problem.
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
# A better way is to run compile and link separately
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed
</pre>


[[File:GgplotPalette.svg|300px]]
Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
<pre>
export R_HOME=/home/brb/Downloads/R-3.0.2
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
./embed # No need to include R CMD in front.
</pre>


== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].
See [[Ggplot2|ggplot2]]


== Data Manipulation & Tidyverse ==
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
* [https://www.tidyverse.org/ Tidyverse] Homepage
* [https://www.rstudio.com/resources/webinars/pipelines-for-data-analysis-in-r/ Pipelines for data analysis in R], [https://www.rstudio.com/resources/videos/data-science-in-the-tidyverse/ Data Science in the Tidyverse]
: <syntaxhighlight lang='bash'>
  Import
    |
    | readr, readxl
    | haven, DBI, httr  +----- Visualize ------+
    |                    |    ggplot2, ggvis    |
    |                    |                      |
  Tidy ------------- Transform
  tibble              dplyr                  Model
  tidyr                  |                    broom
                          +------ Model ---------+
</syntaxhighlight>
* [https://github.com/matloff/TidyverseSkeptic TidyverseSkeptic] by Norm Matloff
* [http://r4ds.had.co.nz/ R for Data Science] and [http://tidyverse.org/ tidyverse] package (it is a collection of '''ggplot2, tibble, tidyr, readr, purrr''' & '''dplyr''' packages).
** tidyverse, among others, was used at [http://juliasilge.com/blog/Mining-CRAN-DESCRIPTION/ Mining CRAN DESCRIPTION Files] (tbl_df(), %>%, summarise(), count(), mutate(), arrange(), unite(), ggplot(), filter(), select(), ...). Note that there is a problem to reproduce the result. I need to run ''cran <- cran[, -14]'' to remove the MD5sum column.
** [http://brettklamer.com/diversions/statistical/compile-r-for-data-science-to-a-pdf/ Compile R for Data Science to a PDF]
* [https://www.rstudio.com/wp-content/uploads/2015/02/data-wrangling-cheatsheet.pdf Data Wrangling with dplyr and tidyr Cheat Sheet]
* [https://hbctraining.github.io/Intro-to-R/lessons/07_intro_tidyverse.html Data Wrangling with Tidyverse] from the Harvard Chan School of Public Health.
* [http://datascienceplus.com/best-packages-for-data-manipulation-in-r/ Best packages for data manipulation in R]. It demonstrates to perform the same tasks using '''data.table''' and '''dplyr''' packages. '''data.table''' is faster and it may be a go-to package when performance and memory are the constraints.
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by Alboukadel Kassambara
** subset data frame columns: pull(), select(), select_if(), other helper functions
** subset (filter) data frame rows: slice(), filter(), filter_all(), filter_if(), filter_at(), sample_n(), top_n()
** identify and remove duplicate rows: duplicated(), unique(), distinct()
** ordering rows: arrange(), desc()
** renaming and adding columns: rename()
** compute and add new variables to a data frame: mutate(), transmutate()
** computing summary statistics (pay to view)


=== Install on Ubuntu ===
=== Create a Simple Socket Server in R ===
[https://stackoverflow.com/a/46983233 How to install Tidyverse on Ubuntu 16.04 and 17.04]
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].  
<syntaxhighlight lang='bash'>
# Ubuntu >= 18.04. However, I get unmet dependencies errors on R 3.5.3.
# r-cran-curl : Depends: r-api-3.4
sudo apt-get install r-cran-curl r-cran-openssl r-cran-xml2


# Works fine on Ubuntu 16.04, 18.04
Create an R function
sudo apt install libcurl4-openssl-dev libssl-dev libxml2-dev
<pre>
</syntaxhighlight>
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>
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 '^]'.


80 R packages will be installed after ''tidyverse'' has been installed.
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                 


=== Install on Raspberry Pi/(ARM based) Chromebook ===
R> quit
In additional to the requirements of installing on Ubuntu, I got an error when it is installing a dependent package [https://github.com/r-lib/fs/issues/146 fs]: '''undefined symbol: pthread_atfork'''. The [https://cran.r-project.org/web/packages/fs/index.html fs] package version is 1.2.6. The [https://github.com/r-lib/fs/issues/128#issuecomment-435552967 solution] is to add one line in fs/src/Makevars file and then install the "fs" package using the source on the local machine.
Connection closed by foreign host.
</pre>


=== [http://rpubs.com/danmirman/Rgroup-part1 5 most useful data manipulation functions] ===
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
* subset() for making subsets of data (natch)
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]].
* merge() for combining data sets in a smart and easy way
* '''melt()'''-reshape2 package for converting from wide to long data formats. See an example [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others here] where we want to combine multiple columns of values into 1 column. melt() is replaced by gather().
* dcast()-reshape2 package for converting from long to wide data formats (or just use [https://datascienceplus.com/building-barplots-with-error-bars/ tapply()]), and for making summary tables
* ddply()-plyr package for doing split-apply-combine operations, which covers a huge swath of the most tricky data operations


=== [https://cran.r-project.org/web/packages/data.table/index.html data.table] ===
See my [[Rserve]] page.
Fast aggregation of large data (e.g. 100GB in RAM or just several GB size file), fast ordered joins, fast add/modify/delete of columns by group using no copies at all, list columns and a fast file reader (fread).


Some resources:
=== outsider ===
* https://www.rdocumentation.org/packages/data.table/versions/1.12.0
* [https://joss.theoj.org/papers/10.21105/joss.02038 outsider]: Install and run programs, outside of R, inside of R
* [https://www.waldrn.com/dplyr-vs-data-table/ R Packages: dplyr vs data.table]
* [https://github.com/stephenturner/om..bcftools Run bcftools with outsider in R]
* [https://github.com/rstudio/cheatsheets/raw/master/datatable.pdf Cheat sheet] from [https://www.rstudio.com/resources/cheatsheets/ RStudio]
* [https://www.r-bloggers.com/importing-data-into-r-part-two/ Reading large data tables in R]. fread(FILENAME)
* Note that '''x[, 2]'' always return 2. If you want to do the thing you want, use ''x[, 2, with=FALSE]'' or ''x[, V2]'' where V2 is the header name. See the FAQ #1 in [http://datatable.r-forge.r-project.org/datatable-faq.pdf data.table].
* [http://r-norberg.blogspot.com/2016/06/understanding-datatable-rolling-joins.html Understanding data.table Rolling Joins]
* [https://rollingyours.wordpress.com/2016/06/14/fast-aggregation-of-large-data-with-the-data-table-package/ Intro to The data.table Package]
** Subsetting rows and/or columns
** Alternative to using tapply(), aggregate(), table() to summarize data
** Similarities to SQL, DT[i, j, by]
* [https://www.listendata.com/2016/10/r-data-table.html R : data.table (with 50 examples)] from ListenData
** Describe Data
** Selecting or Keeping Columns
** Rename Variables
** Subsetting Rows / Filtering
** Faster Data Manipulation with Indexing
** Performance Comparison
** Sorting Data
** Adding Columns (Calculation on rows)
** How to write Sub Queries (like SQL)
** Summarize or Aggregate Columns
** GROUP BY (Within Group Calculation)
** Remove Duplicates
** Extract values within a group
** SQL's RANK OVER PARTITION
** Cumulative SUM by GROUP
** Lag and Lead
** Between and LIKE Operator
** Merging / Joins
** Convert a data.table to data.frame
* [https://www.dezyre.com/data-science-in-r-programming-tutorial/r-data-table-tutorial R Tutorial: data.table] from dezyre.com
** Syntax: DT[where, select|update|do, by]
** Keys and setkey()
** Fast grouping using j and by: DT[,sum(v),by=x]
** Fast ordered joins: X[Y,roll=TRUE]
* In the [https://cran.r-project.org/web/packages/data.table/vignettes/datatable-intro-vignette.html Introduction to data.table] vignette, the data.table::order() function is SLOWER than base::order() from my Odroid xu4 (running Ubuntu 14.04.4 trusty on uSD) <syntaxhighlight lang='rsplus'>
odt = data.table(col=sample(1e7))
(t1 <- system.time(ans1 <- odt[base::order(col)]))  ## uses order from base R
#  user  system elapsed
#  2.730  0.210  2.947
(t2 <- system.time(ans2 <- odt[order(col)]))        ## uses data.table's order
#  user  system elapsed
#  2.830  0.215  3.052
(identical(ans1, ans2))
# [1] TRUE
</syntaxhighlight>
* [https://jangorecki.github.io/blog/2016-06-30/Boost-Your-Data-Munging-with-R.html Boost Your Data Munging with R]
* [https://www.rdocumentation.org/packages/data.table/versions/1.12.0/topics/rbindlist rbindlist()]. One problem, it uses too much memory. In fact, when I try to analyze R package downloads, the command "dat <- rbindlist(logs)" uses up my 64GB memory (OS becomes unresponsive).


[https://github.com/Rdatatable/data.table/wiki/Installation#openmp-enabled-compiler-for-mac OpenMP enabled compiler for Mac]. This instruction works on my Mac El Capitan (10.11.6) when I need to upgrade the data.table version from 1.11.4 to 1.11.6.
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===


Question: how to make use multicore with data.table package?
=== [http://rdotnet.codeplex.com/ R.NET] ===


=== reshape & reshape2 (superceded by tidyr package) ===
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
* [http://r-exercises.com/2016/07/06/data-shape-transformation-with-reshape/ Data Shape Transformation With Reshape()]
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
* Use '''acast()''' function in reshape2 package. It will convert data.frame used for analysis to a table-like data.frame good for display.
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.
* http://lamages.blogspot.com/2013/10/creating-matrix-from-long-dataframe.html


=== [http://cran.r-project.org/web/packages/tidyr/index.html tidyr] and benchmark ===
Terminal
An evolution of reshape2. It's designed specifically for data tidying (not general reshaping or aggregating) and works well with dplyr data pipelines.
{{Pre}}
# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk
</pre>
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
<pre>
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
</pre>
* And then run '''sudo ldconfig'''


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


Make wide tables long with '''gather()''' (see 6.3.1 of Efficient R Programming)
If above does not work, a simple way is by (under Ubuntu) running
<syntaxhighlight lang='rsplus'>
<pre>
library(tidyr)
sudo apt-get install r-cran-rjava
library(efficient)
</pre>
data(pew) # wide table
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.
dim(pew) # 18 x 10,  (religion, '<$10k', '$10--20k', '$20--30k', ..., '>150k')  
pewt <- gather(data = pew, key = Income, value = Count, -religion)
dim(pew) # 162 x 3,  (religion, Income, Count)


args(gather)
=== RCaller ===
# function(data, key, value, ..., na.rm = FALSE, convert = FALSE, factor_key = FALSE)
</syntaxhighlight>
where the three arguments of gather() requires:
* data: a data frame in which column names will become row vaues
* key: the name of the categorical variable into which the column names in the original datasets are converted.
* value: the name of cell value columns


In this example, the 'religion' column will not be included (-religion).
=== RApache ===
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf


=== dplyr, plyr packages ===
=== Rscript, arguments and commandArgs() ===
* plyr package suffered from being slow in some cases. dplyr addresses this by porting much of the computation to C++. Another additional feature is the ability to work with data stored directly in an external '''database'''. The benefits of doing this are the data can be managed natively in a relational database, queries can be conducted on that database, and only the results of query returned.
[https://www.r-bloggers.com/passing-arguments-to-an-r-script-from-command-lines/ Passing arguments to an R script from command lines]
* Essential functions: 3 rows functions, 3 column functions and 1 mixed function.
Syntax:
: <syntaxhighlight lang='rsplus'>
<pre>
          select, mutate, rename
$ Rscript --help
            +------------------+
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
filter      +                  +
</pre>
arrange    +                  +
group_by    +                  +
drop_na    +                  +
ungroup    + summarise        +
            +------------------+
</syntaxhighlight>
* These functions works on data frames and tibble objects. Note stats package also has a filter() function for time series data. If we have not loaded the dplyr package, the filter() function below will give an error (count() also is from dplyr).
: <syntaxhighlight lang='rsplus'>
iris %>% filter(Species == "setosa") %>% count()
head(iris %>% filter(Species == "setosa") %>% arrange(Sepal.Length))
</syntaxhighlight>
* [http://genomicsclass.github.io/book/pages/dplyr_tutorial.html dplyr tutorial] from PH525x series (Biomedical Data Science by Rafael Irizarry and Michael Love). For select() function, some additional options to select columns based on a specific criteria include
** start_with()/ ends_with() = Select columns that start/end with a character string
** contains() = Select columns that contain a character string
** matches() = Select columns that match a regular expression
** one_of() = Select columns names that are from a group of names
* [http://r4ds.had.co.nz/transform.html Data Transformation] in the book '''R for Data Science'''. Five key functions in the '''dplyr''' package:
** Filter rows: filter()
** Arrange rows: arrange()
** Select columns: select()
** Add new variables: mutate()
** Grouped summaries: group_by() & summarise()
: <syntaxhighlight lang='rsplus'>
# filter
jan1 <- filter(flights, month == 1, day == 1)
filter(flights, month == 11 | month == 12)
filter(flights, arr_delay <= 120, dep_delay <= 120)
df <- tibble(x = c(1, NA, 3))
filter(df, x > 1)
filter(df, is.na(x) | x > 1)


# arrange
Example:
arrange(flights, year, month, day)
<pre>
arrange(flights, desc(arr_delay))
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")
</pre>
<pre>
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt
# args[2] =  out.txt
</pre>


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


# mutate
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
flights_sml <- select(flights,
Provides hash-bang (#!) capability for R
  year:day,
  ends_with("delay"),
  distance,
  air_time
)
mutate(flights_sml,
  gain = arr_delay - dep_delay,
  speed = distance / air_time * 60
)
# if you only want to keep the new variables
transmute(flights,
  gain = arr_delay - dep_delay,
  hours = air_time / 60,
  gain_per_hour = gain / hours
)


# summarise()
FAQs:
by_day <- group_by(flights, year, month, day)
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
summarise(by_day, delay = mean(dep_delay, na.rm = TRUE))
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
{{Pre}}
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"


# pipe. Note summarise() can return more than 1 variable.
-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R       # text, R --help
delays <- flights %>%
                                              # Example: R -q -e "date()"
  group_by(dest) %>%
  summarise(
    count = n(),
    dist = mean(distance, na.rm = TRUE),
    delay = mean(arr_delay, na.rm = TRUE)
  ) %>%
  filter(count > 20, dest != "HNL")
flights %>%
  group_by(year, month, day) %>%
  summarise(mean = mean(dep_delay, na.rm = TRUE))
</syntaxhighlight>
* Videos
** [https://youtu.be/jWjqLW-u3hc Hands-on dplyr tutorial for faster data manipulation in R] by Data School. At time 17:00, it compares the '''%>%''' operator, '''with()''' and '''aggregate()''' for finding group mean.
** https://youtu.be/aywFompr1F4 (shorter video) by Roger Peng
** https://youtu.be/8SGif63VW6E by Hadley Wickham
** [https://www.rstudio.com/resources/videos/tidy-eval-programming-with-dplyr-tidyr-and-ggplot2/ Tidy eval: Programming with dplyr, tidyr, and ggplot2]. Bang bang "!!" operator was introduced for use in a function call.
* [https://csgillespie.github.io/efficientR/data-carpentry.html#dplyr Efficient R Programming]
* [http://www.r-exercises.com/2017/07/19/data-wrangling-transforming-23/ Data wrangling: Transformation] from R-exercises.
* [https://rollingyours.wordpress.com/2016/06/29/express-intro-to-dplyr/ Express Intro to dplyr] by rollingyours.
* [https://martinsbioblogg.wordpress.com/2017/05/21/using-r-when-using-do-in-dplyr-dont-forget-the-dot/ the dot].
* [http://martinsbioblogg.wordpress.com/2013/03/24/using-r-reading-tables-that-need-a-little-cleaning/ stringr and plyr] A '''data.frame''' is pretty much a list of vectors, so we use plyr to apply over the list and stringr to search and replace in the vectors.
* https://randomjohn.github.io/r-maps-with-census-data/ dplyr and stringr are used
* [https://datascienceplus.com/5-interesting-subtle-insights-from-ted-videos-data-analysis-in-r/ 5 interesting subtle insights from TED videos data analysis in R]
* [https://www.mango-solutions.com/blog/what-is-tidy-eval-and-why-should-i-care What is tidy eval and why should I care?]


=== stringr ===
-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
* https://www.rstudio.com/wp-content/uploads/2016/09/RegExCheatsheet.pdf
                                              # It won't show the startup message when it is used in the command line.
* [https://github.com/rstudio/cheatsheets/raw/master/strings.pdf stringr Cheat sheet] (2 pages, this will immediately download the pdf file)
                                              # Example: Rscript -e "date()"
* [https://csgillespie.github.io/efficientR/data-carpentry.html#regular-expressions Efficient data carpentry &#8594; Regular expressions] from Efficient R programming book by Gillespie & Lovelace.
</pre>


=== [https://github.com/smbache/magrittr magrittr] ===
We can install littler using two ways.
* [https://cran.r-project.org/web/packages/magrittr/vignettes/magrittr.html Vignettes]
* 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.
* [http://www.win-vector.com/blog/2018/04/magrittr-and-wrapr-pipes-in-r-an-examination/ magrittr and wrapr Pipes in R, an Examination]. Instead of nested statements, it is using pipe operator '''%>%'''. So the code is easier to read. Impressive!
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.
: <syntaxhighlight lang='rsplus'>
x %>% f    # f(x)
x %>% f(y)  # f(x, y)
x %>% f(arg=y)  # f(x, arg=y)
x %>% f(z, .) # f(z, x)
x %>% f(y) %>% g(z)  #  g(f(x, y), z)


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


iris %>%
'''r''' was not meant to run interactively like '''R'''. See ''man r''.
`[[`("Species")


iris %>%
=== RInside: Embed R in C++ ===
`[[`(5)
See [[R#RInside|RInside]]


iris %>%
(''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.
  subset(select = "Species")
</syntaxhighlight>
* '''Split-apply-combine''': group + summarize + sort/arrange + top n. The following example is from [https://csgillespie.github.io/efficientR/data-carpentry.html#data-aggregation Efficient R programming].
: <syntaxhighlight lang='rsplus'>
data(wb_ineq, package = "efficient")
wb_ineq %>%
  filter(grepl("g", Country)) %>%
  group_by(Year) %>%
  summarise(gini = mean(gini, na.rm  = TRUE)) %>%
  arrange(desc(gini)) %>%
  top_n(n = 5)
</syntaxhighlight>
* [https://drdoane.com/writing-pipe-friendly-functions/ Writing Pipe-friendly Functions]
* http://rud.is/b/2015/02/04/a-step-to-the-right-in-r-assignments/
* http://rpubs.com/tjmahr/pipelines_2015
* http://danielmarcelino.com/i-loved-this-crosstable/
* http://moderndata.plot.ly/using-the-pipe-operator-in-r-with-plotly/
* RMSE
: <syntaxhighlight lang='rsplus'>
f <- function(x) {
  (y - x) %>%
    '^'(2) %>%
    sum %>%
    '/'(length(x)) %>%
    sqrt %>%
    round(2)
}
</syntaxhighlight>
* Videos
** [https://www.rstudio.com/resources/videos/writing-readable-code-with-pipes/ Writing Readable Code with Pipes]
** [https://youtu.be/iIBTI_qiq9g Pipes in R - An Introduction to magrittr package]
: <syntaxhighlight lang='rsplus'>
# Examples from R for Data Science-Import, Tidy, Transform, Visualize, and Model
diamonds <- ggplot2::diamonds
diamonds2 <- diamonds %>% dplyr::mutate(price_per_carat = price / carat)


pryr::object_size(diamonds)
The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
pryr::object_size(diamonds2)
pryr::object_size(diamonds, diamonds2)


rnorm(100) %>% matrix(ncol = 2) %>% plot() %>% str()
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'.
rnorm(100) %>% matrix(ncol = 2) %T>% plot() %>% str() # 'tee' pipe
    # %T>% works like %>% except that it returns the lefthand side (rnorm(100) %>% matrix(ncol = 2)) 
    # instead of the righthand side.


# If a function does not have a data frame based api, you can use %$%.
To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
# It explodes out the variables in a data frame.
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>
mtcars %$% cor(disp, mpg)


# For assignment, magrittr provides the %<>% operator
The real build process looks like (check <Makefile> for completeness)
mtcars <- mtcars %>% transform(cyl = cyl * 2) # can be simplified by
<pre>
mtcars %<>% transform(cyl = cyl * 2)
g++ -I/home/brb/Downloads/R-3.0.2/include \
</syntaxhighlight>
    -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>


Upsides of using magrittr: no need to create intermediate objects, code is easy to read.
Hello World example of embedding R in C++.
<pre>
#include <RInside.h>                    // for the embedded R via RInside


When not to use the pipe
int main(int argc, char *argv[]) {
* your pipes are longer than (say) 10 steps
* you have multiple inputs or outputs
* Functions that use the current environment: assign(), get(), load()
* Functions that use lazy evaluation: tryCatch(), try()


=== outer() ===
    RInside R(argc, argv);              // create an embedded R instance


=== Genomic sequence ===
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
* chartr
<syntaxhighlight lang='bash'>
> yourSeq <- "AAAACCCGGGTTTNNN"
> chartr("ACGT", "TGCA", yourSeq)
[1] "TTTTGGGCCCAAANNN"
</syntaxhighlight>


=== broom ===
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
[https://cran.r-project.org/web/packages/broom/index.html broom]: Convert Statistical Analysis Objects into Tidy Tibbles


=== lobstr package - dig into the internal representation and structure of R objects ===
    exit(0);
[https://www.tidyverse.org/articles/2018/12/lobstr/ lobstr 1.0.0]
}
</pre>


== Data Science ==
The above can be compared to the Hello world example in Qt.
See [[Data_science|Data science]] page
<pre>
#include <QApplication.h>
#include <QPushButton.h>


== microbenchmark & rbenchmark ==
int main( int argc, char **argv )
* [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]
    QApplication app( argc, argv );
* [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] ==
    QPushButton hello( "Hello world!", 0 );
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.
    hello.resize( 100, 30 );


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


== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ==
    return app.exec();
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
}
</pre>


== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ==
=== [http://www.rfortran.org/ RFortran] ===
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
RFortran is an open source project with the following aim:


For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
<pre>
sudo apt-get install libgtk2.0-dev libcairo2-dev
</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].
It works only on Windows platform with Microsoft Visual Studio installed:(


== [http://igraph.org/r/ igraph] ==
== Call R from other languages ==
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]
=== C ===
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


== Identifying dependencies of R functions and scripts ==
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://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
<syntaxhighlight lang='rsplus'>
library(mvbutils)
foodweb(where = "package:batr")


foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace


foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
=== JRI ===
</syntaxhighlight>
http://www.rforge.net/JRI/


== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
=== ryp2 ===
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://rpy.sourceforge.net/rpy2.html


Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
== Create a standalone Rmath library ==
R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of [http://cran.r-project.org/doc/manuals/R-admin.html#The-standalone-Rmath-library R-admin manual].


== Colors ==
Here is my experience based on R 3.0.2 on Windows OS.
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
* http://deanattali.com/blog/colourpicker-package/ Not sure what it is doing
* [http://www.lifehack.org/484519/how-to-choose-the-best-colors-for-your-data-charts How to Choose the Best Colors For Your Data Charts]
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* [http://sape.inf.usi.ch/quick-reference/ggplot2/colour Color names in R]


=== [http://rpubs.com/gaston/colortools colortools] ===
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
Tools that allow users generate color schemes and palettes
Suppose we have downloaded R source code and build R from its source. See [[R#Build_R_from_its_source|Build_R_from_its_source]]. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


=== [https://github.com/daattali/colourpicker colourpicker] ===
=== Use Rmath library in our code ===
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
<pre>
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.


=== [https://cran.r-project.org/web/packages/inlmisc/index.html inlmisc] ===
# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
[https://owi.usgs.gov/blog/tolcolors/ GetTolColors()]. Lots of examples.
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# It is OK to save the cpp file under any directory.


== [https://github.com/kevinushey/rex rex] ==
# Force to link against the static library <libRmath.a>
Friendly Regular Expressions
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
# Force to link against dynamic library <Rmath.dll>
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
 
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
 
<pre>
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.
</pre>
</pre>
 
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
Some issues
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
<pre>
<pre>
cat("abcd",
c:\R>RmathEx1
    # This is my comment
Enter a argument for the normal cdf:
    "defg")
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>
</pre>
will result in
 
Below is the cpp program <RmathEx1.cpp>.
<pre>
<pre>
> tidy_source("clipboard")
//RmathEx1.cpp
Error in base::parse(text = code, srcfile = NULL) :  
#define MATHLIB_STANDALONE
   3:1: unexpected string constant
#include <iostream>
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
#include "Rmath.h"
3: "defg"
 
  ^
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;
}
</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.  
 
<pre>
== Calling R.dll directly ==
cat("abcd"
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.
    ,"defg"  # This is my comment
 
   ,"ghij")
== Create HTML report ==
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].
 
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.
 
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
* [http://gforge.se/2020/07/news-in-htmltable-2-0/ News in htmlTable 2.0]
 
=== [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]
 
=== [https://github.com/crubba/htmltab htmltab] package ===
This package is NOT used to CREATE html report but EXTRACT html table.
 
=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.
 
== Create academic report ==
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.
 
== Create pdf and epub files ==
{{Pre}}
# Idea:
#       knitr        pdflatex
#   rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
</pre>
</pre>
will become
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
<pre>
{{Pre}}
cat("abcd", "defg"  # This is my comment
git clone https://github.com/yihui/knitr-examples.git
, "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>
* <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.


== Download papers ==
Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
{{Pre}}
Search and Download Papers from the bioRxiv Preprint Server
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf


=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
git clone https://github.com/yihui/knitr-examples.git
Interface to the arXiv API
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</pre>


=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
{{Pre}}
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
# Idea:
* https://ropensci.org/technotes/2018/12/14/pdftools-20/
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


== [https://github.com/ColinFay/aside aside]: set it aside ==
library(knitr)
An RStudio addin to run long R commands aside your current session.
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</pre>


== Teaching ==
Convert tex to epub
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion


== Organize R research project ==
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
* [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) ===
* https://rmarkdown.rstudio.com/lesson-7.html
[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]
* 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


== Text to speech ==
== Create Word report ==
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


== Weather data ==
=== Using the power of Word ===
* [https://github.com/ropensci/prism prism] package
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


== logR ==
=== knitr + pandoc ===
https://github.com/jangorecki/logR
* 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


== Progress bar ==
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.
https://github.com/r-lib/progress#readme
<pre>
# Idea:
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</pre>
and then
<pre>
FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
</pre>
Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.


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'.
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>


== cron ==
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://github.com/bnosac/cronr cronR]
* 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


= Different ways of using R =
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!
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
<pre>
knit("example.Rmd")
pandoc("example.md", format="epub")
</pre>


== 10 things R can do that might surprise you ==
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
<pre>
> pandoc("Rmd_to_Epub.md", format="epub")
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
Error in (function (input, format, ext, cfg)  : conversion failed
In addition: Warning message:
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
</pre>


== R call C/C++ ==
=== pander ===
Mainly talks about .C() and .Call().
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:


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.
<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.


* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
# http://johnmacfarlane.net/pandoc/
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
# http://rapporter.github.com/pander/
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
# http://rapporter.github.com/pander/#examples
* 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 ===
=== R2wd ===
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
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.  
* We don't need to include double quotes around the C/Fortran subroutines in .C() or .Fortran()
<pre>
* 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().
> library(R2wd)
* stats example: [https://github.com/wch/r-source/blob/trunk/src/library/stats/NAMESPACE NAMESPACE]
> wdGet()
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
    installstatconnDCOM()


(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
This will download and install the current version of statconnDCOM
<syntaxhighlight lang='rsplus'>
library.dynam("libname", package, lib.loc)
</syntaxhighlight>


=== library.dynam.unload() ===
You will need a working Internet connection
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
because installation needs to download a file.
* 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.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]
  argument is of length zero
</pre>


=== gcc ===
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]


== SEXP ==
=== Convert from pdf to word ===
Some examples from packages
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert


* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
=== rtf ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


== R call Fortran ==
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
Package xtable will produce html output.  
* [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)
{{Pre}}
print(xtable(X), type="html")
</pre>


== Embedding R ==
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.


* 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.
=== officer ===
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
<ul>
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report]  by Friedrich Leisch in 2007.
<li>[https://cran.r-project.org/web/packages/officer/index.html CRAN]. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.  
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl
<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, "")


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


This example can be run by
== Powerpoint ==
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
<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>


Note:  
== PDF manipulation ==
# '''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://github.com/pridiltal/staplr staplr]
# '''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.
== 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]
 
== COM client or server ==


=== An example from Bioconductor workshop ===
=== Client ===
* 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:
[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
Create <embed.c> file
<pre>
#include <Rembedded.h>
#include <Rdefines.h>


static void doSplinesExample();
=== Server ===
int
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]
main(int argc, char *argv[])
{
    Rf_initEmbeddedR(argc, argv);
    doSplinesExample();
    Rf_endEmbeddedR(0);
    return 0;
}
static void
doSplinesExample()
{
    SEXP e, result;
    int errorOccurred;


    // create and evaluate 'library(splines)'
== Use R under proxy ==
    PROTECT(e = lang2(install("library"), mkString("splines")));
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
    if (errorOccurred) {
        // handle error
    }
    UNPROTECT(1);


    // 'options(FALSE)' ...
== RStudio ==
    PROTECT(e = lang2(install("options"), ScalarLogical(0)));
* [https://github.com/rstudio/rstudio Github]
    // ... modified to 'options(example.ask=FALSE)' (this is obscure)
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
    SET_TAG(CDR(e), install("example.ask"));
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]
    R_tryEval(e, R_GlobalEnv, NULL);
    UNPROTECT(1);


    // 'example("ns")'
=== rstudio.cloud ===
    PROTECT(e = lang2(install("example"), mkString("ns")));
https://rstudio.cloud/
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
    UNPROTECT(1);
}
</pre>
Then build the executable. Note that I don't need to create R_HOME variable.
<pre>
cd
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest


nano embed.c
=== Launch RStudio ===
# Using a single line will give an error and cannot not show the real problem.
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
 
# A better way is to run compile and link separately
=== Create .Rproj file ===
gcc -I../../include -c embed.c
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.
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed
</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].
With an RStudio project file, you can
<pre>
* Restore .RData into workspace at startup
export R_HOME=/home/brb/Downloads/R-3.0.2
* Save workspace to .RData on exit (or '''save.image'''("Robj.RData") & load("Robj.RData"))
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
* Always save history (even if no saving .RData, '''savehistory'''(".Rhistory") & loadhistory(".Rhistory"))
./embed # No need to include R CMD in front.
* etc
</pre>


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


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


=== Create a Simple Socket Server in R ===
== Visual Studio ==
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].
[http://blog.revolutionanalytics.com/2017/05/r-and-python-support-now-built-in-to-visual-studio-2017.html R and Python support now built in to Visual Studio 2017]


Create an R function
== 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>
simpleServer <- function(port=6543)
list.files(pattern = "^Something")
{
</pre>
  sock <- socketConnection ( port=port , server=TRUE)
 
  on.exit(close( sock ))
Using '''Sys.glob()"' as
  cat("\nWelcome to R!\nR>" ,file=sock )
<pre>
  while(( line <- readLines ( sock , n=1)) != "quit")
> Sys.glob("~/Downloads/*.txt")
  {
[1] "/home/brb/Downloads/ip.txt"       "/home/brb/Downloads/valgrind.txt"
    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
 
== Hidden tool: rsync in Rtools ==
<pre>
<pre>
$ telnet localhost 6543
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
Trying 127.0.0.1...
sending incremental file list
Connected to localhost.
a.exe
Escape character is '^]'.


Welcome to R!
sent 323142 bytes received 31 bytes 646346.00 bytes/sec
R> summary(iris[, 3:5])
total size is 1198416 speedup is 3.71
  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
c:\Rtools\bin>
Connection closed by foreign host.
</pre>
</pre>


=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
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].
Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See [[R#An_example_from_Bioconductor_workshop|Example from Bioconductor workshop]].


See my [[Rserve]] page.
== Install rgdal package (geospatial Data) on ubuntu ==
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>


=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
R
{{Pre}}
install.packages("rgdal")
</pre>


=== [http://rdotnet.codeplex.com/ R.NET] ===
== Install sf package ==
 
I got the following error even I have installed some libraries.  
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.
 
Terminal
<syntaxhighlight lang='bash'>
# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk
</syntaxhighlight>
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
<pre>
<pre>
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
checking GDAL version >= 2.0.1... no
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
configure: error: sf is not compatible with GDAL versions below 2.0.1
</pre>
</pre>
* And then run '''sudo ldconfig'''
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


Now go back to R
sudo apt update
<syntaxhighlight lang='rsplus'>
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
install.packages("rJava")
</syntaxhighlight>
Done!


If above does not work, a simple way is by (under Ubuntu) running
sudo apt install libgdal-dev # works on ubuntu 20.04 too
<pre>
                            # no need the previous lines
sudo apt-get install r-cran-rjava
</pre>
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


=== RCaller ===
== Database ==
* 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]


=== RApache ===
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
'''Creating a new database''':
Provides hash-bang (#!) capability for R
{{Pre}}
library(DBI)


FAQs:
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
dbDisconnect(mydb)
* [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]
unlink("my-db.sqlite")
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
<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
# temporary database
                                              # Example: R -q -e "date()"
mydb <- dbConnect(RSQLite::SQLite(), "")
dbDisconnect(mydb)
</pre>


-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
'''Loading data''':
                                              # It won't show the startup message when it is used in the command line.
{{Pre}}
                                              # Example: Rscript -e "date()"
mydb <- dbConnect(RSQLite::SQLite(), "")
</syntaxhighlight>
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)


We can install littler using two ways.
dbListTables(mydb)
* 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.
dbListFields(con, "mtcars")


'''r''' was not meant to run interactively like '''R'''. See ''man r''.
dbReadTable(con, "mtcars")
</pre>


=== RInside: Embed R in C++ ===
'''Queries''':
See [[R#RInside|RInside]]
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


(''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.
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')


The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))


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


To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
'''Batched queries''':
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>
{{Pre}}
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


The real build process looks like (check <Makefile> for completeness)
dbClearResult(rs)
<pre>
g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include  \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0
</pre>
</pre>


Hello World example of embedding R in C++.
'''Multiple parameterised queries''':
<pre>
{{Pre}}
#include <RInside.h>                    // for the embedded R via RInside
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
 
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
int main(int argc, char *argv[]) {
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)
</pre>


    RInside R(argc, argv);              // create an embedded R instance
'''Statements''':
{{Pre}}
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
#> [1] 0
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</pre>


    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]


    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===


    exit(0);
=== [[MySQL#Use_through_R|RMySQL]] ===
}
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
</pre>
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.


The above can be compared to the Hello world example in Qt.
=== MongoDB ===
<pre>
* http://www.r-bloggers.com/r-and-mongodb/
#include <QApplication.h>
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/
#include <QPushButton.h>


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


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


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


    return app.exec();
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
}
* To use databases with dplyr, you need to first install dbplyr
</pre>
* 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


=== [http://www.rfortran.org/ RFortran] ===
'''Create a new SQLite database''':
RFortran is an open source project with the following aim:
{{Pre}}
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")


''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)


It works only on Windows platform with Microsoft Visual Studio installed:(
copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db
</pre>


== Call R from other languages ==
'''Connect to a database''':
=== C ===
{{Pre}}
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


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]
library(dbplyr)
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")
</pre>


Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
'''Querying the database with the SQL syntax''':
{{Pre}}
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
</pre>


=== JRI ===
'''Querying the database with the dplyr syntax''':
http://www.rforge.net/JRI/
{{Pre}}
surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)


=== ryp2 ===
show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
http://rpy.sourceforge.net/rpy2.html
</pre>


== Create a standalone Rmath library ==
'''Simple database queries''':
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].
{{Pre}}
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</pre>
 
'''Laziness''' (instruct R to stop being lazy):
{{Pre}}
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</pre>
 
'''Complex database queries''':
{{Pre}}
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table


Here is my experience based on R 3.0.2 on Windows OS.
surveys # The plot_id column also features in the surveys table


=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
# Join databases method 1
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.
plots %>%
<pre>
  filter(plot_id == 1) %>%
cd C:\R\R-3.0.2\src\nmath\standalone
  inner_join(surveys) %>%
make -f Makefile.win
  collect()
</pre>
</pre>


=== Use Rmath library in our code ===
=== NoSQL ===
<pre>
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
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
== Github ==
# 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>
=== R source  ===
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
# 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
</pre>
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
<pre>
c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>


Below is the cpp program <RmathEx1.cpp>.
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
<pre>
//RmathEx1.cpp
#define MATHLIB_STANDALONE
#include <iostream>
#include "Rmath.h"


using std::cout; using std::cin; using std::endl;
=== R packages (only) source (metacran) ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.


int main()
=== Bioconductor packages source ===
{
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
  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 << ") = " <<
=== Send local repository to Github in R by using reports package ===
    pnorm(x1, 0, 1, 1, 0)  << endl;
http://www.youtube.com/watch?v=WdOI_-aZV0Y
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
</pre>


== Calling R.dll directly ==
=== My collection ===
See Chapter 8.2.2 of [http://cran.r-project.org/doc/manuals/R-exts.html#Calling-R_002edll-directly|Writing R Extensions]. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.
* https://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 HTML report ==
=== How to download ===
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor.


=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
Clone ~ Download.
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.
* Command line
<pre>
git clone https://gist.github.com/4484270.git
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
* Within R
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
<pre>
library(devtools)
source_gist("4484270")
</pre>
or
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
library(RJSONIO)
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})
</pre>


=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
=== Jekyll ===
* https://github.com/renkun-ken/formattable
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
* 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 ===
== Connect R with Arduino ==
This package is NOT used to CREATE html report but EXTRACT html table.
* 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


=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
== Android App ==
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://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]


== Create academic report ==
== Common plots tips ==
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.
=== Create an empty plot ===
'''plot.new()'''   


== Create pdf and epub files ==
=== Overlay plots ===
<syntaxhighlight lang='rsplus'>
[https://finnstats.com/index.php/2021/08/15/how-to-overlay-plots-in-r/ How to Overlay Plots in R-Quick Guide with Example].
# Idea:
<pre>
#        knitr        pdflatex
#Step1:-create scatterplot
#  rnw -------> tex ----------> pdf
plot(x1, y1)
library(knitr)
#Step 2:-overlay line plot
knit("example.rnw") # create example.tex file
lines(x2, y2)
</syntaxhighlight>
#Step3:-overlay scatterplot
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
points(x2, y2)
<syntaxhighlight lang='bash'>
</pre>
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.
=== Save the par() and restore it ===
'''Example 1''': Don't use old.par <- par() directly. no.readonly = FALSE by default. * The '''`no.readonly = TRUE`''' argument in the [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par par()] function in R is used to get the full list of graphical parameters '''that can be restored'''.
* When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
* If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
<pre>
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)
</pre>
'''Example 2''': Use it inside a function with the [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/on.exit on.exit(0] function.
<pre>
ex <- function() {
  old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
  on.exit(par(old.par))
  ## ... do lots of par() settings and plots
  ## ...
  invisible() #-- now, par(old.par)  will be executed
}
</pre>
'''Example 3''': It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.
<pre>
ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
</pre>
<pre>
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar
</pre>


Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
=== Grouped boxplots ===
<syntaxhighlight lang='bash'>
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
# Idea:
** 'at' parameter in boxplot() to change the equal spaced boxplots
#        knitr        pandoc
** embed par(mar=) in boxplot()
#  rmd -------> md ----------> pdf
** 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)


git clone https://github.com/yihui/knitr-examples.git
=== [https://www.samruston.co.uk/ Weather Time Line] ===
cd knitr-examples
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].
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</syntaxhighlight>


To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
=== Horizontal bar plot ===
<syntaxhighlight lang='rsplus'>
{{Pre}}
# Idea:
library(ggplot2)
#        knitr        pandoc
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
#  rnw -------> tex ----------> markdown or epub
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change") 
</pre>


library(knitr)
[[:File:Ggplot2bar.svg]]
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.
=== Include bar values in a barplot ===
## First install texlive base and extra packages
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
## sudo apt-get install texlive-latex-base texlive-latex-extra
* [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.
pandoc: Could not find media `figure/SchwederSpjotvoll-1', skipping...
* [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]
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>
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
Use text().  
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion


=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
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].
Create Tables In LaTeX, HTML, Markdown And ReStructuredText


* https://rmarkdown.rstudio.com/lesson-7.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.
* 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 ==
=== Grouped barplots ===
 
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
=== knitr + pandoc ===
{{Pre}}
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
library(ggplot2)
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
# mydata <- data.frame(OUTGRP, INGRP, value)
* http://rmarkdown.rstudio.com/articles_docx.html
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
 
      geom_bar(position="dodge", stat="identity")
It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.
<pre>
# Idea:
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</pre>
</pre>
and then
* 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>
{{Pre}}
FILE <- "example"
> 1 - 2*(1-pnorm(1))
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</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.
* [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)


Another way is
=== Math expression ===
<pre>
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
library(pander)
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
name = "demo"
* Some cases
knit(paste0(name, ".Rmd"), encoding = "utf-8")
** Use [https://www.rdocumentation.org/packages/base/versions/3.6.0/topics/expression expression()] function
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
** Don't need the backslash; use ''eta'' instead of ''\eta''. ''eta'' will be recognized as a special keyword in expression()
</pre>
** Use parentheses instead of curly braces; use ''hat(eta)'' instead of ''hat{eta}''
** Summary: use expression(hat(eta)) instead of expression(\hat{\eta})
** [] means subscript, while ^ means superscript. See [https://statisticsglobe.com/add-subscript-and-superscript-to-plot-in-r Add Subscript and Superscript to Plot in R]
** Spacing can be done with ~.
** Mix math symbols and text using paste()
** Using substitute() and paste() if we need to substitute text (this part is advanced)
{{Pre}}
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")
 
# Superscript
plot(1:10, main = expression("My Title"^2))  
# Subscript
plot(1:10, main = expression("My Title"[2])


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:
# Expressions with Spacing
* A pdf file: pandoc -s report.md -t latex -o report.pdf
# '~' is to add space and '*' is to squish characters together
* A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
plot(1:10, xlab= expression(Delta * 'C'))
* Openoffice: pandoc report.md -o report.odt
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
* Word docx: pandoc report.md -o report.docx
    ylab = expression(phi^{rho + a} * z * w),
    main = "Pure Expressions with Spacing")


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!
# Expressions with Text
<pre>
plot(x,y,
knit("example.Rmd")
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
pandoc("example.md", format="epub")
    ylab = expression(paste("Here is some text of ", phi^{rho})),
</pre>
    main = "Expressions with Text")


PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
# Substituting Expressions
<pre>
plot(x,y,
> pandoc("Rmd_to_Epub.md", format="epub")
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
    ylab = substitute(paste("e is = ", e ), list(e = ee)),  
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
    main = "Substituted Expressions")
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>
</pre>


=== pander ===
=== Impose a line to a scatter plot ===
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:
* abline + lsfit # least squares
 
{{Pre}}
<pre>
plot(cars)
library(pander)
abline(lsfit(cars[, 1], cars[, 2]))
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
# OR
            output = tempfile(), convert = 'docx')
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>
</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/
=== How to actually make a quality scatterplot in R: axis(), mtext() ===
# http://rapporter.github.com/pander/
[https://www.r-bloggers.com/2021/08/how-to-actually-make-a-quality-scatterplot-in-r/ How to actually make a quality scatterplot in R]
# http://rapporter.github.com/pander/#examples


=== R2wd ===
=== 3D scatterplot ===
Use [http://cran.r-project.org/web/packages/R2wd/ R2wd] package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.  
* [http://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'''.
<pre>
* [[R_web#plotly|R web > plotly]]
> 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
=== Rotating x axis labels for barplot ===
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
{{Pre}}
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</pre>


You will need a working Internet connection
=== Set R plots x axis to show at y=0 ===
because installation needs to download a file.
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
{{Pre}}
  argument is of length zero
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</pre>
</pre>


The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
=== Different colors of axis labels in barplot ===
See [https://stackoverflow.com/questions/18839731/vary-colors-of-axis-labels-in-r-based-on-another-variable Vary colors of axis labels in R based on another variable]
 
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
{{Pre}}
tN <- table(Ni <- stats::rpois(100, lambda = 5))
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
</pre>


=== Convert from pdf to word ===
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert
{{Pre}}
barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
</pre>


=== rtf ===
=== Use text() to draw labels on X/Y-axis including rotation ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
* adj = 1 means top/right alignment.  For left-bottom alignment, set adj = 0. The default is to center the text. [[https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text ?text]
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
** par("usr") is determined *after* a plot has been created
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
* https://datascienceplus.com/building-barplots-with-error-bars/
{{Pre}}
par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</pre>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/


=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
=== Vertically stacked plots with the same x axis ===
Package xtable will produce html output. <syntaxhighlight lang='rsplus'>print(xtable(X), type="html")</syntaxhighlight>
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in


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.
=== Include labels on the top axis/margin: axis() and mtext() ===
<pre>
plot(1:4, rnorm(4), axes = FALSE)
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3
</pre>
See also [[#15_Questions_All_R_Users_Have_About_Plots| 15_Questions_All_R_Users_Have_About_Plots]]


=== [http://cran.r-project.org/web/packages/ReporteRs/index.html ReporteRs] ===
This can be used to annotate each plot with the script name, date, ...
Microsoft Word, Microsoft Powerpoint and HTML documents generation from R. The source code is hosted on https://github.com/davidgohel/ReporteRs
<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>


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


== PDF manipulation ==
=== Legend tips ===
[https://github.com/pridiltal/staplr staplr]
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]


== R Graphs Gallery ==
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
* [https://www.facebook.com/pages/R-Graph-Gallery/169231589826661 Romain François]
{{Pre}}
* [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].
plot(rnorm(100))
* 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.
# op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)
</pre>


== COM client or server ==
'''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>


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


[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
'''Add a legend title'''
<pre>
legend(, title = "")
</pre>


=== Server ===
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


== Use R under proxy ==
=== Superimpose a density plot or any curves ===
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
Use '''lines()'''.  


== RStudio ==
Example 1
* [https://github.com/rstudio/rstudio Github]
{{Pre}}
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
plot(cars, main = "Stopping Distance versus Speed")
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]
lines(stats::lowess(cars))


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


=== Launch RStudio ===
Example 2
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
{{Pre}}
 
require(survival)
=== Create .Rproj file ===
n = 10000
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.
beta1 = 2; beta2 = -1
 
lambdaT = 1 # baseline hazard
With an RStudio project file, you can
lambdaC = # hazard of censoring
* Restore .RData into workspace at startup
set.seed(1234)
* Save workspace to .RData on exit
x1 = rnorm(n,0)
* Always save history (even if no saving .RData)
x2 = rnorm(n,0)
* etc
# 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>


=== package search ===
Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".
https://github.com/RhoInc/CRANsearcher


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


== Visual Studio ==
[[:File:Logscale.png]]
[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 ==
=== Custom scales ===
* Extension
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
list.files(pattern = "^Something")
</pre>


Using '''Sys.glob()"' as
== Time series ==
<pre>
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
> Sys.glob("~/Downloads/*.txt")
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>


== Hidden tool: rsync in Rtools ==
=== Time series stock price plot ===
<pre>
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
* [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]
sending incremental file list
* https://timelyportfolio.github.io/rCharts_time_series/history.html
a.exe


sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
{{Pre}}
total size is 1198416  speedup is 3.71
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')


c:\Rtools\bin>
tail(Cl(DJI))
</pre>
</pre>
And rsync works best when we need to sync folder.
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
binary/
binary/Eula.txt
binary/cherrytree.lnk
binary/depends64.chm
binary/depends64.dll
binary/depends64.exe
binary/mtputty.exe
binary/procexp.chm
binary/procexp.exe
binary/pscp.exe
binary/putty.exe
binary/sqlite3.exe
binary/wget.exe


sent 4115294 bytes  received 244 bytes  1175868.00 bytes/sec
=== tidyquant: Getting stock data ===
total size is 8036311  speedup is 1.95
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]


c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
=== Timeline plot ===
cygwin warning:
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
  MS-DOS style path detected: c:\users\limingc\Documents\binary\procexp.exe
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R
  Preferred POSIX equivalent is: /cygdrive/c/users/limingc/Documents/binary/procexp.exe
  CYGWIN environment variable option "nodosfilewarning" turns off this warning.
  Consult the user's guide for more details about POSIX paths:
    http://cygwin.com/cygwin-ug-net/using.html#using-pathnames


c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
=== Clockify ===
sending incremental file list
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]
binary/
binary/procexp.exe


sent 1767277 bytes  received 35 bytes  3534624.00 bytes/sec
== Circular plot ==
total size is 8036311  speedup is 4.55
* 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.


c:\Rtools\bin>
== Word cloud ==
</pre>
* [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]


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
== Text mining ==
* [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].


== Install rgdal package (geospatial Data) on ubuntu ==
== World map ==
Terminal
[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)
<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
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
<syntaxhighlight lang='rsplus'>
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]
install.packages("rgdal")
</syntaxhighlight>


== Install sf package ==
=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
I got the following error even I have installed some libraries.  
* [https://blog.rstudio.com/2015/05/01/rstudio-v0-99-preview-graphviz-and-diagrammer/ Graphviz and DiagrammeR]
<pre>
* http://rich-iannone.github.io/DiagrammeR/,
checking GDAL version >= 2.0.1... no
** [http://rich-iannone.github.io/DiagrammeR/io.html#r-markdown rmarkdown]
configure: error: sf is not compatible with GDAL versions below 2.0.1
** [http://rich-iannone.github.io/DiagrammeR/graphviz_and_mermaid.html graphviz and mermaid] doc and examples
</pre>
* https://donlelek.github.io/2015-03-31-dags-with-r/
Then I follow the instruction here
* [https://mikeyharper.uk/flowcharts-in-r-using-diagrammer/ Data-driven flowcharts in R using DiagrammeR]
<syntaxhighlight lang='bash'>
 
sudo apt remove libgdal-dev
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
sudo apt remove libproj-dev
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable


sudo apt update
=== DAGitty (browser-based and R package) ===
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
* http://dagitty.net/
sudo apt install libgdal-dev
* https://cran.r-project.org/web/packages/dagitty/index.html
</syntaxhighlight>


== Set up Emacs on Windows ==
=== dagR ===
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
* https://cran.r-project.org/web/packages/dagR
<pre>
(setq-default inferior-R-program-name
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>


== Database ==
=== Gmisc ===
* https://cran.r-project.org/web/views/Databases.html
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]
* [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] ===
=== Concept Maps ===
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.
* https://github.com/rstats-db/RSQLite


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


mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
== Venn Diagram ==
dbDisconnect(mydb)
[[Venn_diagram|Venn diagram]]
unlink("my-db.sqlite")


# temporary database
== hexbin plot ==
mydb <- dbConnect(RSQLite::SQLite(), "")
* [https://datasciencetut.com/how-to-create-a-hexbin-chart-in-r/ How to create a hexbin chart in R]
dbDisconnect(mydb)
* [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.
</syntaxhighlight>


'''Loading data''':
== Bump chart/Metro map ==
<syntaxhighlight lang='rsplus'>
https://dominikkoch.github.io/Bump-Chart/
mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)


dbListTables(mydb)
== Amazing/special plots ==
See [[Amazing_plot|Amazing plot]].


dbListFields(con, "mtcars")
== Google Analytics ==
=== GAR package ===
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html


dbReadTable(con, "mtcars")
== Linear Programming ==
</syntaxhighlight>
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/


'''Queries''':
== Linear Algebra ==
<syntaxhighlight lang='rsplus'>
* [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.
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')
* [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.


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
== Amazon Alexa ==
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
== R and Singularity ==
https://rviews.rstudio.com/2017/03/29/r-and-singularity/


res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
== Teach kids about R with Minecraft ==
dbFetch(res)
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html
</syntaxhighlight>


'''Batched queries''':
== Secure API keys ==
<syntaxhighlight lang='rsplus'>
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


dbClearResult(rs)
== Credentials and secrets ==
</syntaxhighlight>
[https://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]


'''Multiple parameterised queries''':
== Hide a password ==
<syntaxhighlight lang='rsplus'>
=== keyring package ===
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
* https://cran.r-project.org/web/packages/keyring/index.html
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
* [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]
nrow(dbFetch(rs))
 
#> [1] 4
=== getPass ===
dbClearResult(rs)
[https://cran.r-project.org/web/packages/getPass/README.html getPass]
</syntaxhighlight>
 
== Vision and image recognition ==
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image


'''Statements''':
== Creating a Dataset from an Image ==
<syntaxhighlight lang='rsplus'>
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]
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)
</syntaxhighlight>


=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
== Turn pictures into coloring pages ==
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0


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


=== [[MySQL#Use_through_R|RMySQL]] ===
* [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]
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.


=== MongoDB ===
== Ryacas: R Interface to the 'Yacas' Computer Algebra System ==
* http://www.r-bloggers.com/r-and-mongodb/
[https://blog.ephorie.de/doing-maths-symbolically-r-as-a-computer-algebra-system-cas Doing Maths Symbolically: R as a Computer Algebra System (CAS)]
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


=== odbc ===
== Game ==
* [https://kbroman.org/miner_book/?s=09 R Programming with Minecraft]
* [https://cran.r-project.org/web/packages/pixelpuzzle/index.html pixelpuzzle]
* [https://www.rostrum.blog/2022/09/24/pixeltrix/ Interactive pixel art in R with {pixeltrix}]
* [https://rtaoist.blogspot.com/2021/03/r-shiny-maths-games-for-6-years-old.html Shiny math game]
* [https://cran.microsoft.com/web/packages/mazing/index.html mazing]: Utilities for Making and Plotting Mazes
* [https://github.com/jeroenjanssens/raylibr/blob/main/demo/snake.R snake] which is based on [https://github.com/jeroenjanssens/raylibr raylibr]


=== RODBC ===
== Music ==
* [https://flujoo.github.io/gm/ gm]. Require to install [https://musescore.org/en MuseScore], an open source and free notation software.


=== DBI ===
== SAS ==
[https://github.com/MangoTheCat/sasMap sasMap] Static code analysis for SAS scripts


=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
= R packages =
* To use databases with dplyr, you need to first install dbplyr
[[R_packages|R packages]]
* 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


'''Create a new SQLite database''':
= Tricks =
<syntaxhighlight lang='rsplus'>
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")


my_db_file <- "portal-database.sqlite"
== Getting help ==
my_db <- src_sqlite(my_db_file, create = TRUE)
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.
* https://stat.ethz.ch/pipermail/r-help/
* https://stat.ethz.ch/pipermail/r-devel/


copy_to(my_db, surveys)
== Better Coder/coding, best practices ==
copy_to(my_db, plots)
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
my_db
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
</syntaxhighlight>
* [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


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


library(dbplyr)
== Getting user's home directory ==
library(dplyr)
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?]
mammals <- src_sqlite("portal_mammals.sqlite")
{{Pre}}
</syntaxhighlight>
# Windows
normalizePath("~")   # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")   # "C:/Users/brb/Documents"


'''Querying the database with the SQL syntax''':
# Mac
<syntaxhighlight lang='rsplus'>
normalizePath("~")  # [1] "/Users/brb"
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
Sys.getenv("R_USER") # [1] ""
</syntaxhighlight>
Sys.getenv("HOME")   # "/Users/brb"


'''Querying the database with the dplyr syntax''':
# Linux
<syntaxhighlight lang='rsplus'>
normalizePath("~")   # [1] "/home/brb"
surveys <- tbl(mammals, "surveys")
Sys.getenv("R_USER") # [1] ""
surveys %>%
Sys.getenv("HOME")   # [1] "/home/brb"
    select(year, species_id, plot_id)
</pre>
head(surveys, n = 10)


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


'''Simple database queries''':
== Distinguish Windows and Linux/Mac, R.Version() ==
<syntaxhighlight lang='rsplus'>
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</syntaxhighlight>


'''Laziness''' (instruct R to stop being lazy):
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
data_subset <- surveys %>%
<pre>
   filter(weight < 5) %>%
get_os <- function(){
   select(species_id, sex, weight) %>%
  sysinf <- Sys.info()
  collect()
  if (!is.null(sysinf)){
</syntaxhighlight>
    os <- sysinf['sysname']
    if (os == 'Darwin')
      os <- "osx"
   } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
   tolower(os)
}
</pre>
<pre>
names(R.Version())
#  [1] "platform"      "arch"          "os"            "system"       
#  [5] "status"        "major"          "minor"          "year"         
#  [9] "month"          "day"            "svn rev"        "language"     
# [13] "version.string" "nickname"
getRversion()
# [1] ‘4.3.0’
</pre>


'''Complex database queries''':
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
<syntaxhighlight lang='rsplus'>
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
plots <- tbl(mammals, "plots")
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
plots # # The plot_id column features in the plots table
* 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]


surveys # The plot_id column also features in the surveys table
If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
<pre>
R_LIBS_SITE=F:/R/library
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''. In R, run '''Sys.getenv("R_LIBS_SITE")''' or '''Sys.getenv("R_LIBS_USER")''' to query the environment variable. See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/EnvVar.html Environment Variables].


# Join databases method 1
=== What is the best place to save Rconsole on Windows platform ===
plots %>%
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()
</syntaxhighlight>


=== NoSQL ===
My preferred settings:
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
* 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


== Github ==
and others (default options)
 
* pagebg: white
=== R source  ===
* pagetext: navy
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
* highlight: DarkRed
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].


=== R packages (only) source (metacran) ===
=== How R starts up ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.
https://rstats.wtf/r-startup.html


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


=== Send local repository to Github in R by using reports package ===
== Saving and loading history automatically: .Rprofile & local() ==
http://www.youtube.com/watch?v=WdOI_-aZV0Y
<ul>
<li>[http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html savehistory("filename")]. It will save everything from the beginning to the command savehistory() to a text file.
<li>'''.Rprofile''' will automatically be loaded when R has started from that directory
<li>Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
<li>'''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
<li>[http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
* You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
* At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
options(continue="  ") # default is "+ "
options(prompt="R> ", continue=" ")
options(editor="nano") # default is "vi" on Linux
# options(htmlhelp=TRUE)
 
local({r <- getOption("repos")
      r["CRAN"] <- "https://cran.rstudio.com"
      options(repos=r)})


=== My collection ===
.First <- function(){
* https://github.com/arraytools
# library(tidyverse)
* https://gist.github.com/4383351 heatmap using leukemia data
cat("\nWelcome at", date(), "\n")
* https://gist.github.com/4382774 heatmap using sequential data
}
* https://gist.github.com/4484270 biocLite


=== How to download ===
.Last <- function(){
cat("\nGoodbye at ", date(), "\n")
</pre>
<li>https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
<li>The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
<li>[https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
</ul>
'''Linux''' or '''Mac'''


Clone ~ Download.  
In '''~/.profile''' or '''~/.bashrc''' I put:
* Command line
<pre>
export R_HISTFILE=~/.Rhistory
</pre>
In '''~/.Rprofile''' I put:
<pre>
<pre>
git clone https://gist.github.com/4484270.git
if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


* Within R
'''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.
<pre>
<pre>
library(devtools)
if (interactive()) {
source_gist("4484270")
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>
</pre>
or
 
First download the json file from
== Disable "Save workspace image?" prompt when exit R? ==
https://api.github.com/users/MYUSERLOGIN/gists
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in R?]
and then
 
== R release versions ==
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
 
== getRversion() ==
<pre>
<pre>
library(RJSONIO)
getRversion()
x <- fromJSON("~/Downloads/gists.json")
[1] ‘4.3.0’
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>


=== Jekyll ===
== Detect number of running R instances in Windows ==
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
* 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.


== Connect R with Arduino ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
* http://bio7.org/?p=2049
* http://www.rforge.net/Arduino/svn.html


== Android App ==
Image Name                    PID Session Name        Session#    Mem Usage
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
============================================================================
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
Rgui.exe                      1096 Console                    1    44,712 K
* [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 ==
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
=== Grouped boxplots ===
* [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] ===
Image Name                    PID Session Name        Session#    Mem Usage
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].
============================================================================
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"


=== Horizontal bar plot ===
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
<syntaxhighlight lang='rsplus'>
</pre>
library(ggplot2)
 
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
== Editor ==
                        "DDR", "BUM", "MAT", "HED", "EXP"),
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change")  
</syntaxhighlight>


[[File:Ggplot2bar.svg|300px]]
<ul>
<li>Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
<pre>
(setq-default inferior-R-program-name
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>
* [https://blog.rwhitedwarf.com/post/use_emacs_for_r/ Using Emacs for R] 2022
</ul>
* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager. The new version (0.98) also provides a new feature for debugging step-by-step. See also [https://www.rstudio.com/rviews/2016/11/11/easy-tricks-you-mightve-missed/ RStudio Tricks]
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code. RStudio can also do this (see the bottom of the code panel).
* [http://rgedit.sourceforge.net/ Rgedit] which includes a feature of splitting screen into two panes and run R in the bottom panel. See [http://www.stattler.com/article/using-gedit-or-rgedit-r here].
* Komodo IDE with browser preview http://www.youtube.com/watch?v=wv89OOw9roI at 4:06 and http://docs.activestate.com/komodo/4.4/editor.html


=== Include bar values in a barplot ===
== GUI for Data Analysis ==
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023
* [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().  
=== 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]


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].
=== 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.


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.
=== Deducer ===
http://cran.r-project.org/web/packages/Deducer/index.html


=== Grouped barplots ===
=== jamovi ===
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)<syntaxhighlight lang='rsplus'>
* https://www.jamovi.org/
library(ggplot2)
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
# 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 ===
== Scope ==
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
See
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
* https://andyphilips.github.io/blog/2017/08/16/mathematical-symbols-in-r-plots.html
** Use [https://www.rdocumentation.org/packages/base/versions/3.6.0/topics/expression expression()] function
** Don't need the backslash; use ''eta'' instead of ''\eta''. ''eta'' will be recognized as a special keyword in expression()
** Use parentheses instead of curly braces; use ''hat(eta)'' instead of ''hat{eta}''
** Summary: use expression(hat(eta)) instead of expression(\hat{\eta})
** [] means subscript, while ^ means superscript
** Spacing can be done with ~.
** Mix math symbols and text using paste()
** Using substitute() and paste() if we need to substitute text (this part is advanced)
: <syntaxhighlight lang='rsplus'>
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")


# Expressions with Spacing
=== source() ===
# '~' is to add space and '*' is to squish characters together
* [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.
plot(1:10, xlab= expression(Delta * 'C'))
* [[#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()''')
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
{{Pre}}
plot(x,y,  
## foo.R ##
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
cat(ArrayTools, "\n")
    ylab = expression(paste("Here is some text of ", phi^{rho})),
## End of foo.R
    main = "Expressions with Text")
 
# 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


# Substituting Expressions
# 2. OK. Make the variable global
plot(x,y,
predict <- function() {
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
  ArrayTools <<- "C:/Program Files'
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
  source("foo.R")
    main = "Substituted Expressions")
}
</syntaxhighlight>
predict()
ArrayTools


=== Impose a line to a scatter plot ===
# 3. OK. Create a global variable
* abline + lsfit # least squares
ArrayTools <- "C:/Program Files"
: <syntaxhighlight lang='rsplus'>
predict <- function() {
plot(cars)
  source("foo.R")
abline(lsfit(cars[, 1], cars[, 2]))
}
# OR
predict()
abline(lm(cars[,2] ~ cars[,1]))
</pre>
</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 ===
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
<syntaxhighlight lang='rsplus'>
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</syntaxhighlight>


=== Different colors of axis labels in barplot ===
Example 1.
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]
<pre>
> 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
</pre>


Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
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.
<syntaxhighlight lang='rsplus'>
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")
</syntaxhighlight>


Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
<syntaxhighlight lang='rsplus'>
barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
</syntaxhighlight>


=== Use [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text text()] to draw labels on X/Y-axis including rotation ===
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
* adj = 1 means top/rigth alignment. The default is to center the text.
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
** par("usr") is determined *after* a plot has been created
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
* https://datascienceplus.com/building-barplots-with-error-bars/
<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 ===
=== New environment ===
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in
* http://adv-r.had.co.nz/Environments.html.
* [https://www.r-bloggers.com/2011/06/environments-in-r/ Environments in R]
* load(), attach(), with().
* [https://stackoverflow.com/questions/33109379/how-to-switch-to-a-new-environment-and-stick-into-it How to switch to a new environment and stick into it?] seems not possible!


=== Increase/decrease legend font size ===
Run the same function on a bunch of R objects
https://stackoverflow.com/a/36842578
{{Pre}}
<syntaxhighlight lang='rsplus'>
mye = new.env()
plot(rnorm(100))
load(<filename>, mye)
op <- par(cex=2)
for(n in names(mye)) n = as_tibble(<nowiki>mye[[n]]</nowiki>)
legend("topleft", legend = 1:4, col=1:4, pch=1)
</pre>
par(op)
</syntaxhighlight>


=== Superimpose a density plot or any curves ===
Just look at the contents of rda file without saving to anywhere (?load)
Use '''lines()'''.  
<pre>
 
local({
Example 1
  load("myfile.rda")
<syntaxhighlight lang='rsplus'>
  ls()
plot(cars, main = "Stopping Distance versus Speed")
})
lines(stats::lowess(cars))
</pre>
</syntaxhighlight>
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>
 
=== 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 [https://www.rdocumentation.org/packages/utils/versions/3.6.1/topics/str ?str]


Example 2
If we use str() on a function like str(lm), it is equivalent to args(lm)
<syntaxhighlight lang='rsplus'>
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
</syntaxhighlight>


=== Custom scales ===
For a complicated list object, it is useful to use the '''max.level''' argument; e.g. str(, max.level = 1)
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]


== Time series ==
For a large data frame, we can use the '''tibble()''' function; e.g. mydf %>% tibble()
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]


=== Time series stock price plot ===
=== tidy() function ===
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See [[Tidyverse#broom|here]].
* [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'>
=== View all objects present in a package, ls() ===
library(quantmod)
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
getSymbols("AAPL")
{{Pre}}
getSymbols("IBM") # similar to AAPL
> devtools::load_all("mypackage")
getSymbols("CSCO") # much smaller than AAPL, IBM
> search()
getSymbols("DJI") # Dow Jones, huge
[1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
[4] "package:stats"    "package:graphics"  "package:grDevices"
     col='orange', subset = '2017::2017-08')
[7] "package:utils"    "package:datasets"  "package:methods"
[10] "Autoloads"        "package:base"


tail(Cl(DJI))
> ls("package:mypackage")
</syntaxhighlight>
[1] "_mypackage_rcpp_hello_world" "evalCpp"                    "library.dynam.unload"     
[4] "rcpp_hello_world"            "system.file"
</pre>


=== Timeline plot ===
Note that the first argument of ls() (or detach()) is used to specify the environment. It can be
https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* 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).


== Circular plot ==
== Speedup R code ==
* http://freakonometrics.hypotheses.org/20667 which uses https://cran.r-project.org/web/packages/circlize/ circlize] package.
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
* 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 ==
=== Profiler ===
* [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.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler] (Video)
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
* [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]
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]


== World map ==
== && vs & ==
[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)
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  


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


=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
<pre>
* http://rich-iannone.github.io/DiagrammeR/
c(T,F,T) & c(T,T,T)
* https://donlelek.github.io/2015-03-31-dags-with-r/
# [1]  TRUE FALSE  TRUE
 
c(T,F,T) && c(T,T,T)
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
# [1] TRUE
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
</pre>
<pre>
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated
</pre>
It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.
<pre>
> foo <- function(arg, L) {
  # Suppose 'L' is meaningful only if 'arg' is provided
  #
  # Evaluate 'L' only if 'arg' is provided
  #
  if (!missing(arg) && L) {
    print("L is true")
  } else {
    print("Either arg is missing or L is FALSE")
  }
}
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"
</pre>
Other examples: '''&&''' is more flexible than '''&'''.
<pre>
nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))


=== DAGitty (browser-based and R package) ===
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
* http://dagitty.net/
</pre>
* https://cran.r-project.org/web/packages/dagitty/index.html


=== dagR ===
== for-loop, control flow ==
* https://cran.r-project.org/web/packages/dagR
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Control ?Control]
* '''next''' can be used to skip the rest of the inner-most loop
* [https://www.programiz.com/r/ifelse-function ifelse() Function]


== Venn Diagram ==
== Vectorization ==
* limma http://www.ats.ucla.edu/stat/r/faq/venn.htm - only black and white?
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
* VennDiagram - input has to be the numbers instead of the original vector?
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* 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]
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
# systemPipeR package method
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
library(systemPipeR)
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
OLlist <- overLapper(setlist[1:3], type="vennsets")
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
vennPlot(list(OLlist))                           
* [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].


# R script source method
=== sapply vs vectorization ===
source("http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R")
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
# or (obtained by dput(setlist))
setlist <- structure(list(A = c("o", "h", "u", "p", "i", "s", "a", "w",
"b", "z", "n", "c", "k", "j", "y", "m", "t", "q"), B = c("h",
"r", "x", "y", "b", "t", "d", "o", "m", "q", "g", "v", "c", "u",
"f", "z"), C = c("b", "e", "t", "u", "s", "j", "o", "k", "d",
"l", "g", "i", "w", "n", "p", "a", "y", "x", "m", "z"), D = c("f",
"g", "b", "k", "j", "m", "e", "q", "i", "d", "o", "l", "c", "t",
"x", "r", "s", "u", "w", "a", "z", "n"), E = c("u", "w", "o",
"k", "n", "h", "p", "z", "l", "m", "r", "d", "q", "s", "x", "b",
"v", "t"), F = c("o", "j", "r", "c", "l", "l", "u", "b", "f",
"d", "u", "m", "y", "t", "y", "s", "a", "g", "t", "m", "x", "m"
)), .Names = c("A", "B", "C", "D", "E", "F"))


OLlist <- overLapper(setlist[1:3], type="vennsets")
=== lapply vs for loop ===
counts <- list(sapply(OLlist$Venn_List, length)) 
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
vennPlot(counts=counts)                         
* https://code-examples.net/en/q/286e03a
</syntaxhighlight>
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]


[[File:Vennplot.png|250px]]
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
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
{{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


== Bump chart/Metro map ==
# bigmemory vignette
https://dominikkoch.github.io/Bump-Chart/
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                    function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
</pre>
* Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
{{Pre}}
split(mtcars,mtcars$cyl)


== Amazing plots ==
split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
=== New R logo 2/11/2016 ===
split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))  # split.data.frame() works for matrices
* http://rud.is/b/2016/02/11/plot-the-new-svg-r-logo-with-ggplot2/
</pre>
* https://www.stat.auckland.ac.nz/~paul/Reports/Rlogo/Rlogo.html
* Split columns of a data frame/matrix.  
<syntaxhighlight lang='rsplus'>
{{Pre}}
library(sp)
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
library(maptools)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
library(ggplot2)
# $`1`
library(ggthemes)
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
# rgeos requires the installation of GEOS from http://trac.osgeo.org/geos/
#
system("curl http://download.osgeo.org/geos/geos-3.5.0.tar.bz2 | tar jx")
# $`2`
system("cd geos-3.5.0; ./configure; make; sudo make install")
#  [1] 16  9  4  1  0  1  4  9 16 25
library(rgeos)
</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.
r_wkt_gist_file <- "https://gist.githubusercontent.com/hrbrmstr/07d0ccf14c2ff109f55a/raw/db274a39b8f024468f8550d7aeaabb83c576f7ef/rlogo.wkt"
* 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()].  
if (!file.exists("rlogo.wkt")) download.file(r_wkt_gist_file, "rlogo.wkt")
{{Pre}}
rlogo <- readWKT(paste0(readLines("rlogo.wkt", warn=FALSE))) # rgeos
tSsp = split.default(testStat, lls)  
# testStat is a vector of numerics including probeset IDs as names
rlogo_shp <- SpatialPolygonsDataFrame(rlogo, data.frame(poly=c("halo", "r"))) # sp
# lls is a vector of entrez IDs (same length as testStat)
rlogo_poly <- fortify(rlogo_shp, region="poly") # ggplot2
# tSSp is a list of the same length as unique elements of lls.
ggplot(rlogo_poly) +
  geom_polygon(aes(x=long, y=lat, group=id, fill=id)) +
  scale_fill_manual(values=c(halo="#b8babf", r="#1e63b5")) +
  coord_equal() +
  theme_map() +
  theme(legend.position="none")
</syntaxhighlight>


=== 3D plot ===
sapply(tSsp, function(x) names(which.max(x)))
Using [https://chitchatr.wordpress.com/2010/06/28/fun-with-persp-function/ persp] function to create the following plot. Code in [https://gist.github.com/arraytools/ef955d017cb6b9ef0690a4fe79f809f9 github].
# return a vector of probset IDs of length of unique entrez IDs
</pre>


[[File:3dpersp.png|200px]]
=== strsplit and sapply ===
{{Pre}}
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
> strsplit(namedf, " ")
[[1]]
[1] "John" "ABC"


=== Christmas tree ===
[[2]]
http://wiekvoet.blogspot.com/2014/12/merry-christmas.html. Code in [https://gist.github.com/arraytools/668404a33d32a6652d4dddf5d294689e github].
[1] "Mary" "CDE"


[[File:XMastree.png|150px]]
[[3]]
[1] "Kat" "FGH"


=== Happy Thanksgiving ===
> sapply(strsplit(namedf, " "), "[", 1)
[http://blog.revolutionanalytics.com/2015/11/happy-thanksgiving.html Turkey]
[1] "John" "Mary" "Kat"
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"
</pre>


[[File:Turkey.png|150px]]
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
<ul>
<li>[https://stackoverflow.com/questions/35925529/reduce-columns-of-a-matrix-by-a-function-in-r Reduce columns of a matrix by a function in R]. To use rowMedians() instead of rowMeans(), we need to install [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats] from CRAN.
<syntaxhighlight lang='r'>
set.seed(1)
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res  # notice the sorting of columns
      a  b  c
[1,] 46  1 31
[2,] 47 12 32
[3,] 48 13 33
[4,] 49 14 34
[5,] 50 15 35
[6,] 51 16 36
[7,] 52 17 37
[8,] 53 18 38
[9,] 54 19 39
[10,] 55 20 40


=== Happy Valentine's Day ===
# vapply() is safter than sapply().
* [https://rud.is/b/2017/02/14/geom%E2%9D%A4%EF%B8%8F/ Geom❤️] 2017
# The 3rd arg in vapply() is a template of the return value.
* [http://www.theanalyticslab.nl/2019/02/14/nerds-on-valentines-day/ Happy Valentines day by Nerds] 2019
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


=== treemap ===
apply(x, 1, mean, na.rm=T)
http://ipub.com/treemap/
# [1] 31 27 28 29 30 31 32 33 34 35
</pre>
</li>
<li>[https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
</li>
<li>[https://www.statforbiology.com/2020/stat_r_tidyverse_columnwise/ From ''for()'' loops to the ''split-apply-combine'' paradigm for column-wise tasks: the transition for a dinosaur]
</li>
</ul>


[[File:TreemapPop.png|150px]]
=== Mean of duplicated rows: colMeans and rowsum ===
 
<ul>
=== [https://en.wikipedia.org/wiki/Voronoi_diagram Voronoi diagram] ===
<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'''.
* https://www.stat.auckland.ac.nz/~paul/Reports/VoronoiTreemap/voronoiTreeMap.html
{{Pre}}
* http://letstalkdata.com/2014/05/creating-voronoi-diagrams-with-ggplot/
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


=== Silent Night ===
aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
[[File:Silentnight.png|200px]]
#  Group.1  V1  V2  V3  V4  V5  V6
# 1      a 10.0 20.0 30.0 40.0 50.0 60.0
# 2      b  1.5 12.0 22.0 31.5 41.5 51.5
# 3      c  4.0 14.0 24.0 34.0 44.0 54.0
# 4      d  7.5 17.5 27.5 37.5 47.5 57.5
</pre>
<li>[[Arraytools#Reducing_multiple_probes.2Fprobe_sets_to_one_per_gene_symbol|Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays]]


The code in [https://gist.github.com/arraytools/6d841de27bec48fa0b72559e9aeb13ad github].
</li>
<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}}
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
</pre>
</li>
</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.
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
<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)


=== The Travelling Salesman Portrait ===
# Another example: select rows with a minimum value from a certain column (yval in this case)
https://fronkonstin.com/2018/04/04/the-travelling-salesman-portrait/
> 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
</pre>
</li>
</ul>


=== Moon phase calendar ===
=== Mean by Group ===
https://chichacha.netlify.com/2018/05/26/making-calendar-with-ggplot-moon-phase-calendar/
[https://statisticsglobe.com/mean-by-group-in-r Mean by Group in R (2 Examples) | dplyr Package vs. Base R]
<pre>
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].


=== Calendar heatmap ===
== Apply family ==
https://stackoverflow.com/questions/26171068/add-dates-to-calendar-heat-map-r
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.  


=== Chaos ===
The following list gives a hierarchical relationship among these functions.
[https://fronkonstin.com/2019/01/10/rcpp-camaron-de-la-isla-and-the-beauty-of-maths/ Rcpp, Camarón de la Isla and the Beauty of Maths]
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
*** '''replicate'''(n, expr, simplify = "array")
** '''mapply'''(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
*** '''Vectorize'''(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
*** '''Map'''(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
** '''vapply'''(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
** '''rapply'''(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
* '''tapply'''(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a [https://en.wikipedia.org/wiki/Jagged_array "Ragged" Array]. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
** '''aggregate'''(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each '''columns''' of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame. This is used to '''summarize''' data.
** '''by'''(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each '''subset data frame''' split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
* '''eapply'''(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment


== Google Analytics ==
[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?]
=== GAR package ===
* 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://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-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.


== Linear Programming ==
Some short examples:
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
* [http://www.datasciencemadesimple.com/apply-function-r/ Apply Function in R – apply vs lapply vs sapply vs mapply vs tapply vs rapply vs vapply] from datasciencemadesimple.com.
* [https://stackoverflow.com/a/7141669 How to use which one (apply family) when?]


== Linear Algebra ==
=== Apply vs for loop ===
* [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.
Note that, apply's performance is not always better than a for loop. See
* [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.
* 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`?]


== Amazon Alexa ==
=== Progress bar ===
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
[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?]


== R and Singularity ==
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.rstudio.com/rviews/2017/03/29/r-and-singularity/


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


== Secure API keys ==
=== simplify option in sapply() ===
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
<pre>
library(KEGGREST)


== Hide a password: keyring package ==
names1 <- keggGet(c("hsa05340", "hsa05410"))
* https://cran.r-project.org/web/packages/keyring/index.html
names2 <- sapply(names1, function(x) x$GENE)
* [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]
length(names2)  # same if we use lapply() above
# [1] 2


== Vision and image recognition ==
names3 <- keggGet(c("hsa05340"))
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
names4 <- sapply(names3, function(x) x$GENE)
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify
# [1] 1
</pre>


== Turn pictures into coloring pages ==
=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
<ul>
<li>mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
<pre>
mapply(rep, 1:4, 4:1)
mapply(rep, times = 1:4, x = 4:1)
mapply(function(x, y) seq_len(x) + y,
      c(a =  1, b = 2, c = 3),  # names from first
      c(A = 10, B = 0, C = -10))
mapply(c, firstList, secondList, SIMPLIFY=FALSE)
</pre>
</li>
<li>[https://bensstats.wordpress.com/2020/10/06/robservations-3-finding-the-expected-value-of-the-maximum-of-two-bivariate-normal-variables-with-simulation/ Finding the Expected value of the maximum of two Bivariate Normal variables with simulation] sapply + mapply.
<pre>
z <- mapply(function(u, v) { max(u, v) },
            u = x[, 1], v = x[, 2])
</pre>
</li>
<li>[http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming </li>
<li>Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
<ul>
<li>If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply().
{{Pre}}
# Syntax: Map(f, ...)


== Numerical optimization ==
xs <- replicate(5, runif(10), simplify = FALSE)
* [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]
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
Map(weighted.mean, xs, ws)
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares


= R packages =
# instead of a more clumsy way
[[R_packages|R packages]]
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, ...)


= Tricks =
> 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
</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]


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


== Better Coder/coding, best practices ==
See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
* http://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://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
=== rapply - recursive version of lapply ===
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
* http://4dpiecharts.com/tag/recursive/
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


== Getting user's home directory ==
=== replicate ===
See [https://cran.r-project.org/bin/windows/base/rw-FAQ.html#What-are-HOME-and-working-directories_003f What are HOME and working directories?]
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
<syntaxhighlight lang='rsplus'>
{{Pre}}
# Windows
> replicate(5, rnorm(3))
normalizePath("~")  # "C:\\Users\\brb\\Documents"
          [,1]      [,2]      [,3]      [,4]        [,5]
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
Sys.getenv("HOME")  # "C:/Users/brb/Documents"
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</pre>


# Mac
See [[#parallel_package|parSapply()]] for a parallel version of replicate().
normalizePath("~")   # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # "/Users/brb"


# Linux
=== Vectorize ===
normalizePath("~")   # [1] "/home/brb"
* [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()'''.
Sys.getenv("R_USER") # [1] ""
{{Pre}}
Sys.getenv("HOME")   # [1] "/home/brb"
> rep(1:4, 4:1)
</syntaxhighlight>
[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


== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
[[2]]
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
[1] 2 2 2
* 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))
[[3]]
* [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]
[1] 3 3
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]


If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
[[4]]
<pre>
[1] 4
R_LIBS_SITE=F:/R/library
</pre>
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy]  
{{Pre}}
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
[1] 1.6491761 0.9610109
</pre>
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''.
* 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


Note that on Windows OS, R/etc contains
myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
<pre>
myfunc2(1, 2) # 2
$ ls -l /c/Progra~1/r/r-3.2.0/etc
myfunc2(3, 5) # 15
total 142
myfunc2(c(1,3), c(2,5)) # NA
-rw-r--r--    1   Administ    1043 Jun 20  2013 Rcmd_environ
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
-rw-r--r--    1  Administ    1924 Mar 17  2010 Rconsole
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
-rw-r--r--    1   Administ      943 Oct  3 2011 Rdevga
                                        # parameter will be re-used
-rw-r--r--    1   Administ      589 May 20  2013 Rprofile.site
</pre>
-rw-r--r--    1   Administ  251894 Jan 17  2015 curl-ca-bundle.crt
drwxr-xr-x    1  Administ        0 Jun  8 10:30 i386
-rw-r--r--    1  Administ    1160 Dec 31  2014 repositories
-rw-r--r--    1  Administ    30188 Mar 17  2010 rgb.txt
drwxr-xr-x    3  Administ        0 Jun  8 10:30 x64


$ ls /c/Progra~1/r/r-3.2.0/etc/i386
== plyr and dplyr packages ==
Makeconf
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]


$ cat /c/Progra~1/r/r-3.2.0/etc/Rconsole
[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
# 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
[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.
# This can be `yes' (for MDI) or `no' (for SDI).
  MDI = yes
# MDI = no


# the next two are only relevant for MDI
# plyr has a common syntax -- easier to remember
toolbar = yes
# plyr requires less code since it takes care of the input and output format
statusbar = no
# plyr can easily be run in parallel -- faster


## Font.
Tutorials
# Please use only fixed width font.
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
# If font=FixedFont the system fixed font is used; in this case
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
# points and style are ignored. If font begins with "TT ", only
* [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.
# True Type fonts are searched for.
font = TT Courier New
points = 10
style = normal # Style can be normal, bold, italic


# Dimensions (in characters) of the console.
Examples of using dplyr:
rows = 25
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
columns = 80
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
# Dimensions (in characters) of the internal pager.
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
pgrows = 25
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.  
pgcolumns = 80
# should options(width=) be set to the console width?
setwidthonresize = yes
 
# memory limits for the console scrolling buffer, in chars and lines
# NB: bufbytes is in bytes for R < 2.7.0, chars thereafter.
bufbytes = 250000
buflines = 8000


# Initial position of the console (pixels, relative to the workspace for MDI)
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
# xconsole = 0
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
# yconsole = 0


# Dimension of MDI frame in pixels
Tibble objects
# Format (w*h+xorg+yorg) or use -ve w and h for offsets from right bottom
* it does not have row names (cf data frame),
# This will come up maximized if w==0
* it never changes the type of the inputs (e.g. it never converts strings to factors!),
# MDIsize = 0*0+0+0
* it never changes the names of variables
# 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
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
# or in multiple windows (one for each topic)
# pagerstyle can be set to `singlewindow' or `multiplewindows'
pagerstyle = multiplewindows


## Colours for console and pager(s)
{{Pre}}
# (see rwxxxx/etc/rgb.txt for the known colours).
> data(pew, package = "efficient")
background = White
> dim(pew)  
normaltext = NavyBlue
[1] 18 10
usertext = Red
> class(pew) # tibble is also a data frame!!
highlight = DarkRed
[1] "tbl_df"    "tbl"        "data.frame"


## Initial position of the graphics window
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
## (pixels, <0 values from opposite edge)
# A tibble: 162 x 3
xgraphics = -25
                                                      religion Income Count
ygraphics = 0
                                                          <chr>  <chr> <int>
1                                                    Agnostic  <$10k    27
2                                                      Atheist  <$10k    12
...
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
[1] NA
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count,  :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
[1] 181.6975
</pre>


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


## Default setting for console buffering: 'yes' or 'no'
tbObj %>% print(n= nrow(.))
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
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.
bin  COPYING  etc  lib  library  modules  site-library  SVN-REVISION


brb@brb-T3500:~$ ls /usr/lib/R/etc
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].  
javaconf  ldpaths  Makeconf  Renviron  Renviron.orig  Renviron.site  Renviron.ucf  repositories  Rprofile.site
{{Pre}}
TibbleObject$VarName
# OR
TibbleObject[["VarName"]]
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore


brb@brb-T3500:~$ ls /usr/local/lib/R
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
site-library
# OR
dplyr::select(TibbleObject, 2:5) #
</pre>
</pre>
and  
 
'''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]
<pre>
<pre>
brb@brb-T3500:~$ cat /usr/lib/R/etc/Rprofile.site
my_data <- as_tibble(iris)
##                                              Emacs please make this -*- R -*-
class(my_data)
## empty Rprofile.site for R on Debian
</pre>
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Rprofile and Rprofile.site


# ## Example of .Rprofile
To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)
# options(width=65, digits=5)
# 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
=== llply() ===
# local({
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.
#  # add MASS to the default packages, set a CRAN mirror
<pre>
#  old <- getOption("defaultPackages"); r <- getOption("repos")
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
#  r["CRAN"] <- "http://my.local.cran"
</pre>
#  options(defaultPackages = c(old, "MASS"), repos = r)
where rLLID is a list of entrez ID. For example,
#})
<pre>
brb@brb-T3500:~$ cat /usr/lib/R/etc/Renviron.site
get("org.Hs.egGO")[["6772"]]
##                                              Emacs please make this -*- R -*-
</pre>
## empty Renviron.site for R on Debian
returns a list of 49 GOs.
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Renviron and Renviron.site


# ## Example ~/.Renviron on Unix
=== ddply() ===
# R_LIBS=~/R/library
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
# PAGER=/usr/local/bin/less


# ## Example .Renviron on Windows
=== ldply() ===
# R_LIBS=C:/R/library
[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]
# MY_TCLTK="c:/Program Files/Tcl/bin"
 
=== Performance/speed comparison ===
[https://www.r-bloggers.com/2023/01/performance-comparison-of-converting-list-to-data-frame-with-r-language/ Performance comparison of converting list to data.frame with R language]


# ## Example of setting R_DEFAULT_PACKAGES (from R CMD check)
== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
# R_DEFAULT_PACKAGES='utils,grDevices,graphics,stats'
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
# # this loads the packages in the order given, so they appear on
 
# # the search path in reverse order.
=== get_seed() ===
brb@brb-T3500:~$
See the same blog
{{Pre}}
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
</pre>
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== What is the best place to save Rconsole on Windows platform ===
=== Random seeds ===
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See [https://stat.ethz.ch/R-manual/R-patched/library/base/html/Random.html ?Random].
<pre>
set.seed(as.numeric(Sys.time()))


My preferred settings:
set.seed(as.numeric(Sys.Date())) # same seed for each day
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
</pre>
* 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)
=== .Machine and the largest integer, double ===
* pagebg: white
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].
* pagetext: navy
{{Pre}}
* highlight: DarkRed
                          Linux/Mac  32-bit Windows 64-bit Windows
* dataeditbg: white
double.eps              2.220446e-16  2.220446e-16  2.220446e-16
* dataedittext: navy (View() function)
double.neg.eps          1.110223e-16  1.110223e-16  1.110223e-16
* dataedituser: red
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
* editorbg: white (edit() function)
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
* editortext: black
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
</pre>


== Saving and loading history automatically: .Rprofile & local() ==
=== NA when overflow ===
* http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html
* '''.Rprofile''' will automatically be loaded when R has started from that directory
* Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
* '''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
* [http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
** You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
** At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
<pre>
options(continue="  ") # default is "+ "
tmp <- 156287L
options(editor="nano") # default is "vi" on Linux
tmp*tmp
# options(htmlhelp=TRUE)
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647
</pre>


local((r <- getOption("repos")
== How to select a seed for simulation or randomization ==
  r["CRAN"] <- "http://cran.rstudio.com"
* [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]
  options(repos = r)))
* [https://www.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]


.First <- function(){
== set.seed() allow alphanumeric seeds ==
# library(Hmisc)
https://stackoverflow.com/a/10913336
cat("\nWelcome at", date(), "\n")
}


.Last <- function(){
== set.seed(), for loop and saving random seeds ==
cat("\nGoodbye at ", date(), "\n")
<ul>
<li>[https://www.jottr.org/2020/09/21/detect-when-the-random-number-generator-was-used/ Detect When the Random Number Generator Was Used]
</pre>  
* https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
* The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
* [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
 
'''Linux''' or '''Mac'''
 
In '''~/.profile''' or '''~/.bashrc''' I put:
<pre>
<pre>
export R_HISTFILE=~/.Rhistory
if (interactive()) {
  invisible(addTaskCallback(local({
    last <- .GlobalEnv$.Random.seed
   
    function(...) {
      curr <- .GlobalEnv$.Random.seed
      if (!identical(curr, last)) {
        msg <- "NOTE: .Random.seed changed"
        if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
        message(msg)
        last <<- curr
      }
      TRUE
    }
  }), name = "RNG tracker"))
}
</pre>
</pre>
In '''~/.Rprofile''' I put:
</li>
<li>http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
<pre>
<pre>
if (interactive()) {
set.seed(1001)  
  if (.Platform$OS.type == "unix") .First <- function() try(utils::loadhistory("~/.Rhistory"))  
data <- vector("list", 30)  
   .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
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]]
</pre>
</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].


'''Windows'''
== sample() ==
 
=== sample() inaccurate on very large populations, fixed in R 3.6.0 ===
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.
* [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.  
<pre>
{{Pre}}
if (interactive()) {
# R 3.5.3
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
set.seed(123)
}
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1
# 400070 599930
</pre>
</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


== R release versions ==
# R 3.6.0
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
# 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>


== Detect number of running R instances in Windows ==
=== Getting different results with set.seed() in RStudio ===
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
[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().''
<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"
=== dplyr::sample_n() ===
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.


Image Name                    PID Session Name        Session#    Mem Usage
== Regular Expression ==
============================================================================
See [[Regular_expression|here]].
Rgui.exe                      1096 Console                    1    44,712 K


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


Image Name                    PID Session Name        Session#    Mem Usage
== 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.
Rserve.exe                    6108 Console                    1    381,796 K
<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>
</pre>
In R, we can use
<li>File connections
<pre>
<pre>
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
read_chars <- function(file_name) {
[1] ""                                                                          
  conn <- file(file_name, "r")
[2] "Image Name                    PID Session Name        Session#    Mem Usage"
  on.exit(close(conn), add = TRUE)
[3] "============================================================================"
  readChar(conn, file.info(file_name)$size)
[4] "Rgui.exe                      1096 Console                    1    44,804 K"
}
 
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
</pre>
</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>


== Editor ==
== file, connection ==
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
* read() and write()
* read.table() and write.table()
{{Pre}}
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)


* 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).
foo <- function() {
* [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]
  con <- file()
* [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].
  on.exit(close(con))
* 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
  ...
}
</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>


== GUI for Data Analysis ==
=== withr package ===
https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].


=== Rcmdr ===
== Clipboard (?connections), textConnection(), pipe() ==
http://cran.r-project.org/web/packages/Rcmdr/index.html
<ul>
 
<li>On Windows, we can use readClipboard() and writeClipboard().
=== Deducer ===
{{Pre}}
http://cran.r-project.org/web/packages/Deducer/index.html
source("clipboard")
 
read.table("clipboard")
=== jamovi ===
</pre></li>
* https://www.jamovi.org/
<li>Clipboard -> R. Reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function:
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
{{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>
 
=== clipr ===
[https://cran.rstudio.com/web/packages/clipr/ clipr]: Read and Write from the System Clipboard


== Scope ==
== read/manipulate binary data ==
See
* x <- readBin(fn, raw(), file.info(fn)$size)
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
* rawToChar(x[1:16])
* [[#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()''')
* See Biostrings C API


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


# 1. Error
=== format(): padding with zero ===
predict <- function() {
<pre>
  ArrayTools <- "C:/Program Files" # or through load() function
ngenes <- 10
  source("foo.R")                 # or through a function call; foo()
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
}
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
predict()  # Object ArrayTools not found
</pre>


# 2. OK. Make the variable global
=== noquote() ===
predict <- function() {
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict() 
ArrayTools


# 3. OK. Create a global variable
=== stringr package ===
ArrayTools <- "C:/Program Files"
* https://stringr.tidyverse.org/index.html
predict <- function() {
* [https://stringr.tidyverse.org/articles/from-base.html Vignette compares stringr functions to their base R equivalents]
  source("foo.R")
* 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].
}
predict()
</syntaxhighlight>


'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
=== glue package ===
<ul>
<li>[https://cran.r-project.org/web/packages/glue/index.html glue]. Useful in a loop and some function like ggtitle() or ggsave().
<pre>
library(glue)
name <- "Fred"
glue('My name is {name}.')  # My name is Fred.
</pre>
</li>
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
</ul>


Example 1.  
=== Raw data type ===
[https://twitter.com/hadleywickham/status/1387747735441395712 Fun with strings], [https://en.wikipedia.org/wiki/Cyrillic_alphabets Cyrillic alphabets]
<pre>
<pre>
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
a1 <- "А"
> ttt
a2 <- "A"
  type JpnTest
a1 == a2
1   a    999
# [1] FALSE
2    b    999
charToRaw("А")
3    c    999
# [1] d0 90
4    d    999
charToRaw("A")
5    e    999
# [1] 41
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1   a      N5
2    b    999
3    c    999
4    d    999
5    e    999
> ttt
  type JpnTest
1    a    999
2    b    999
3    c    999
4    d    999
5    e    999
</pre>
</pre>


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.
=== number of characters limit ===
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]


Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
== HTTPs connection ==
HTTPS connection becomes default in R 3.2.2. See
* http://blog.rstudio.org/2015/08/17/secure-https-connections-for-r/
* http://blog.revolutionanalytics.com/2015/08/good-advice-for-security-with-r.html


Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
[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)


=== New environment ===
== setInternet2 ==
http://adv-r.had.co.nz/Environments.html
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.


Run the same function on a bunch of R objects
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.
<syntaxhighlight lang='rsplus'>
<pre>
mye = new.env()
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
load(<filename>, mye)
              "GCF_000001405.13.assembly.txt")
for(n in names(mye)) n = as_tibble(mye[[n]])
f1 <- tempfile()
</syntaxhighlight>
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.


=== View all objects present in a package, ls() ===
The following R command will show the exact svn revision for the R you are currently using.
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
<pre>
<syntaxhighlight lang='rsplus'>
R.Version()$"svn rev"
> devtools::load_all("mypackage")
</pre>
> 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")
If setInternet2(T), then https protocol is supported in download.file().  
[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
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.  
* 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 ==
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].
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+


== Profiler ==
'''R up to 3.2.2'''
(Video) [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler]
<pre>
 
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
== && vs & ==
</pre>
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  
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).


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.
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).


== stopifnot(): function argument sanity check ==
'''R 3.3.0'''
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]
<pre>
setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>


== Vectorization ==
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.
* [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].


=== sapply vs vectorization ===
== Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan() ==
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]
In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, '''Inf''' and '''NaN''' as input or output.


=== lapply vs for loop ===
See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance 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?]


=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
[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]
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 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' 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)))
== replace() function ==
# return a vector of probset IDs of length of unique entrez IDs
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)  
</syntaxhighlight>
* https://stackoverflow.com/a/11811147
: 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>


=== Mean of duplicated columns: rowMeans ===
== File/path operations ==
* [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'>
* list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
* file.info()
colnames(x) <- c("A","A", "b", "b", "b", "c"); x
* dir.create()
res <- sapply(split(1:ncol(x), colnames(x)),  
* file.create()
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
* file.copy()
res
* file.exists()
<ul>
<li>'''basename'''() - remove the parent path, '''dirname'''() - returns the part of the path up to but excluding the last path separator
<pre>
> file.path("~", "Downloads")
[1] "~/Downloads"
> dirname(file.path("~", "Downloads"))
[1] "/home/brb"
> basename(file.path("~", "Downloads"))
[1] "Downloads"
</pre>
</li></ul>
* '''path.expand'''("~/.Renviron")  # "/home/brb/.Renviron"
<ul>
<li> '''normalizePath'''() # Express File Paths in Canonical Form
<pre>
> cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
/usr/lib/R
/tmp/RtmpzvDhAe
</pre>
</li>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/system.file system.file()] - Finds the full file names of files in packages etc
<pre>
> system.file("extdata", "ex1.bam", package="Rsamtools")
[1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
</pre>
</li></ul>
* tools::file_path_sans_ext() - [https://stackoverflow.com/a/29114021 remove the file extension] or the sub() function.


# vapply() is safter than sapply().
== read/download/source a file from internet ==
# The 3rd arg in vapply() is a template of the return value.
=== Simple text file http ===
res2 <- vapply(split(1:ncol(x), colnames(x)),
<pre>
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
              rep(0, nrow(x)))
</pre>
</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)
=== Zip, RData, gz file and url() function ===
# [1] 31 27 28 29 30 31 32 33 34 35
<pre>
</syntaxhighlight>
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
</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.


=== Mean of duplicated rows: colMeans and rowsum ===
Another example is [https://stackoverflow.com/a/9548672 Read gzipped csv directly from a url in R]
* colMeans(x, na.rm = FALSE, dims = 1)
<pre>
: <syntaxhighlight lang='rsplus'>
con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
                      "aa_combined-20110321.csv.gz", sep="")))
rownames(x) <- c(rep("a", 2), rep("b", 3), rep("c", 4), "d")
txt <- readLines(con)
res <- sapply(split(1:nrow(x), rownames(x)),
dat <- read.csv(textConnection(txt))
              function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
</pre>
res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
 
</syntaxhighlight>
Another example of using url() is
* rowsum(x, group, reorder = TRUE, …)
<pre>
: <syntaxhighlight lang='rsplus'>
load(url("http:/www.example.com/example.RData"))
x <- matrix(runif(100), ncol = 5) # 20 x 5
</pre>
group <- sample(1:8, 20, TRUE)
 
(xsum <- rowsum(x, group)) # 8 x 5
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].
</syntaxhighlight>
 
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
'''Dropbox''' is easy and works for load(), wget, ...
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
 
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.
* [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]
=== zip function ===
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
This will include 'hallmarkFiles' root folder in the files inside zip.
* [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)
zip(zipfile = 'myFile.zip',
dim(mtcars)
    files = dir('hallmarkFiles', full.names = TRUE))
[1] 32 11
 
> head(mtcars)
# Verify/view the files. 'list = TRUE' won't extract
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
unzip('testZip.zip', list = TRUE)  
Mazda RX4        21.0  6  160 110 3.90 2.620 16.46  0  1    4    4
</pre>
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
=== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ===
Hornet 4 Drive    21.4  6  258 110 3.08 3.215 19.44  1  0    3    1
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.
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
=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
> with(mtcars, table(cyl, vs))
{{Pre}}
  vs
require(RCurl)
cyl  0  1
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
  4  1 10
read.csv(textConnection(myCsv))
  6  3  4
</pre>
  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)
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
> mydf <- read.table(header=T, text='
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into 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
</syntaxhighlight>


== Apply family ==
=== Github files https using RCurl package ===
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.  
* 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


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


[https://www.queryhome.com/tech/76799/r-difference-between-apply-vs-sapply-vs-lapply-vs-tapply Difference between apply vs sapply vs lapply vs tapply?]
=== skimr: A frictionless, pipeable approach to dealing with summary statistics ===
* apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]
* 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:
=== modelsummary ===
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
[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
* [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 ===
=== broom ===
Note that, apply's performance is not always better than a for loop. See
[[Tidyverse#broom|Tidyverse->broom]]
* 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?]
* [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 ===
=== Create publication tables using '''tables''' package ===
[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?]
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]


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


=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
=== fgsea example ===
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
[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]  
* mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming
* Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
** If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply(). <syntaxhighlight lang='rsplus'>
# Syntax: Map(f, ...)


xs <- replicate(5, runif(10), simplify = FALSE)
=== (archived) ClinReport: Statistical Reporting in Clinical Trials ===
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
https://cran.r-project.org/web/packages/ClinReport/index.html
Map(weighted.mean, xs, ws)


# instead of a more clumsy way
== Append figures to PDF files ==
lapply(seq_along(xs), function(i) {
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
  weighted.mean(xs[[i]], ws[[i]])
})
</syntaxhighlight>
** Reduce() reduces a vector, x, to a single value by <span style="color: red">recursively</span> calling a function, f, two arguments at a time. A good example of using '''Reduce()''' function is to read a list of matrix files and merge them. See [https://stackoverflow.com/questions/29820029/how-to-combine-multiple-matrix-frames-into-one-using-r How to combine multiple matrix frames into one using R?] <syntaxhighlight lang='rsplus'>
# Syntax: Reduce(f, x, ...)


> m1 <- data.frame(id=letters[1:4], val=1:4)
== Save base graphics as pseudo-objects ==
> m2 <- data.frame(id=letters[2:6], val=2:6)
[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.
> merge(m1, m2, "id", all = T)
<pre>
  id val.x val.y
pdf(NULL)
1  a    1    NA
dev.control(displaylist="enable")
2 b    2    2
plot(df$x, df$y)
3  c    3    3
text(40, 0, "Random")
4  d    4    4
text(60, 2, "Text")
5  e    NA    5
lines(stats::lowess(df$x, df$y))
6  f    NA    6
p1.base <- recordPlot()
> m <- list(m1, m2)
invisible(dev.off())
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
  id val.x val.y
1  a    1    NA
2  b    2    2
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
</syntaxhighlight>
* [https://statcompute.wordpress.com/2018/09/08/playing-map-and-reduce-in-r-subsetting/ Playing Map() and Reduce() in R – Subsetting] - using parallel and future packages. [https://statcompute.wordpress.com/2018/09/22/union-multiple-data-frames-with-different-column-names/ Union Multiple Data.Frames with Different Column Names]


=== sapply & vapply ===
# Display the saved plot
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
grid::grid.newpage()
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
p1.base
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
</pre>
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]


See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
== Extracting tables from PDFs ==
<ul>
<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)
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
txt <- pdf_text(pdf_file) # length = number of pages
# Suppose the table we are interested in is on page 1
cat(txt[1]) # Good but not in a data frame format


=== rapply - recursive version of lapply ===
pdf_data(pdf_file)[[1]]  # data frame/tibble format
* http://4dpiecharts.com/tag/recursive/
</pre>
* [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].
However, it seems it does not work on [http://www.bloodjournal.org/content/109/8/3177/tab-figures-only Table S6]. Tabulizer package is better at this case.
 
This is another example. [https://mp.weixin.qq.com/s?__biz=MzAxMDkxODM1Ng==&mid=2247490327&idx=1&sn=cca7d4423426318e0c23adb098cf0ad7&chksm=9b485bacac3fd2ba2196b380c59b5eab9d29795d3334b040f50a2fa58124ec6e3be9472829e0&scene=21#wechat_redirect 神技能-自动化批量从PDF里面提取表格]
</li>
<li>[https://www.linuxuprising.com/2019/05/how-to-convert-pdf-to-text-on-linux-gui.html?m=1 How To Convert PDF To Text On Linux (GUI And Command Line)]. It works when I tested my PDF file.
{{Pre}}
sudo apt install poppler-utils
pdftotext -layout input.pdf output.txt
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</pre>
</li>
<li>[https://www.adobe.com/acrobat/how-to/pdf-to-excel-xlsx-converter.html Convert PDF files into Excel spreadsheets] using Adobe Acrobat. See [https://helpx.adobe.com/acrobat/how-to/extract-pages-from-pdf.html How to extract pages from a PDF]. Note the PDF file should not be opened by Excel since it is binary format Excel can't recognize.
<li>I found it is easier to use copy the column (it works) from PDF and paste them to Excel </li>
</ul>
 
== Print tables ==
 
=== addmargins() ===
* [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins]. Puts Arbitrary Margins On Multidimensional Tables Or Arrays.
* [https://datasciencetut.com/how-to-put-margins-on-tables-or-arrays-in-r/ How to put margins on tables or arrays in R?]


=== replicate ===
=== tableone ===
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
* https://cran.r-project.org/web/packages/tableone/
<syntaxhighlight lang='rsplus'>
* [https://datascienceplus.com/table-1-and-the-characteristics-of-study-population/ Table 1 and the Characteristics of Study Population]
> replicate(5, rnorm(3))
* [https://www.jianshu.com/p/e76f2b708d45 如何快速绘制论文的表1(基本特征三线表)?]
          [,1]      [,2]      [,3]      [,4]        [,5]
* See Table 1 from [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</syntaxhighlight>


See [[#parallel_package|parSapply()]] for a parallel version of replicate().
=== 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]


=== Vectorize ===
=== finalfit package ===
* [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/Vectorize Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE)]: creates a function wrapper that vectorizes a scalar function. Its value is a list or vector or array. It calls '''mapply()'''. <syntaxhighlight lang='rsplus'>
[https://finalfit.org/index.html summary_factorlist()] from the finalfit package.
> 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]]
=== table1 ===
[1] 2 2 2
* 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.


[[3]]
=== gtsummary ===
[1] 3 3
* [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.


[[4]]
=== gt ===
[1] 4
[https://www.r-bloggers.com/2024/02/introduction-to-clinical-tables-with-the-gt-package/ Introduction to Clinical Tables with the {gt} Package]
</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 ===
myfunc2(1, 2) # 2
https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.
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 ==
=== tables::tabular() ===
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]


[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
=== gmodels::CrossTable() ===
https://www.statmethods.net/stats/frequencies.html


[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.
=== base::prop.table(x, margin) ===
[http://developer.r-project.org/blosxom.cgi/R-devel/2020/02/13#n2020-02-13 New function ‘proportions()’ and ‘marginSums()’. These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’.] for R 4.0.0.
<pre>
R> m <- matrix(1:4, 2)
R> prop.table(m, 1) # row percentage
          [,1]      [,2]
[1,] 0.2500000 0.7500000
[2,] 0.3333333 0.6666667
R> prop.table(m, 2) # column percentage
          [,1]      [,2]
[1,] 0.3333333 0.4285714
[2,] 0.6666667 0.5714286
</pre>


# plyr has a common syntax -- easier to remember
=== stats::xtabs() ===
# plyr requires less code since it takes care of the input and output format
# plyr can easily be run in parallel -- faster


Tutorials
=== stats::ftable() ===
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
{{Pre}}
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
> ftable(Titanic, row.vars = 1:3)
* [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.
                  Survived  No Yes
 
Class Sex    Age                 
Examples of using dplyr:
1st  Male  Child            0  5
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
            Adult          118  57
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
      Female Child            0  1
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
            Adult            4 140
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.  
2nd  Male  Child            0  11
 
            Adult          154  14
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
      Female Child            0  13
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
            Adult          13  80
 
3rd  Male  Child          35  13
Tibble objects
            Adult          387  75
* it does not have row names (cf data frame),
      Female Child          17  14
* it never changes the type of the inputs (e.g. it never converts strings to factors!),
            Adult          89  76
* it never changes the names of variables
Crew  Male  Child            0  0
 
            Adult          670 192
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
      Female Child            0  0
 
            Adult            3 20
<syntaxhighlight lang='rsplus'>
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
> data(pew, package = "efficient")
            Survived  No Yes
> dim(pew)  
Class Sex                   
[1] 18 10
1st  Male            118  62
> class(pew) # tibble is also a data frame!!
      Female            4 141
[1] "tbl_df"    "tbl"       "data.frame"
2nd  Male            154  25
 
      Female          13  93
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
3rd  Male            422  88
# A tibble: 162 x 3
      Female          106  90
                                                      religion Income Count
Crew  Male            670 192
                                                          <chr> <chr> <int>
      Female            3  20
  1                                                     Agnostic <$10k    27
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
  2                                                     Atheist <$10k    12
            Survived  No Yes
  ...
Sex    Class               
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
Male  1st            118  62
[1] NA
      2nd            154  25
Warning message:
      3rd            422  88
In mean.default(tidyr::gather(pew, key = Income, value = Count, :
      Crew          670 192
  argument is not numeric or logical: returning NA
Female 1st              4 141
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
      2nd            13  93
[1] 181.6975
      3rd            106  90
</syntaxhighlight>
      Crew            3  20
 
> str(Titanic)
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.
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
 
- attr(*, "dimnames")=List of 4
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
  ..$ Class  : chr [1:4] "1st" "2nd" "3rd" "Crew"
<syntaxhighlight lang='rsplus'>
  ..$ Sex     : chr [1:2] "Male" "Female"
TibbleObject$VarName
  ..$ Age    : chr [1:2] "Child" "Adult"
# OR
  ..$ Survived: chr [1:2] "No" "Yes"
TibbleObject[["VarName"]]
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
# OR
> x
pull(TibbleObject, VarName) # won't be a tibble object anymore
          gear  3 4  5
 
cyl vs am             
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
4  0  0        0  0  0
# OR
      1        0  0  1
dplyr::select(TibbleObject, 2:5) #
    1  0        1  2  0
</syntaxhighlight>
      1        0  6 1
 
6  0  0        0  0 0
=== llply() ===
      1       0 2  1
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.
    1 0        2  0
<pre>
      1        0  0  0
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
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
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
</pre>
</pre>
where rLLID is a list of entrez ID. For example,
<pre>
get("org.Hs.egGO")[["6772"]]
</pre>
returns a list of 49 GOs.


=== ddply() ===
== tracemem, data type, copy ==
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]


=== ldply() ===
== Tell if the current R is running in 32-bit or 64-bit mode ==
[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]
<pre>
8 * .Machine$sizeof.pointer
</pre>
where '''sizeof.pointer''' returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.


== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
== 32- and 64-bit ==
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
* Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).


=== get_seed() ===
== Handling length 2^31 and more in R 3.0.0 ==
See the same blog
<syntaxhighlight lang='rsplus'>
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
</syntaxhighlight>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== .Machine and the largest integer, double ===
From R News for 3.0.0 release:
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 ==
''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.
[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]
''


== set.seed() allow alphanumeric seeds ==
In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error
https://stackoverflow.com/a/10913336
<pre>
> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector
</pre>
 
However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
<pre>
> 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
>
</pre>
Note:
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].
# [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17330 [[<- for index 2^31 fails]
 
== 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.


== set.seed(), for loop and saving random seeds ==
* Question: What is TRUE & NA?
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.
Answer: NA


<syntaxhighlight lang='rsplus'>
* Question: What is FALSE & NA?
set.seed(1001)
Answer: FALSE
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]]
</syntaxhighlight>
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.


== sample() inaccurate on very large populations, fixed in R 3.6.0 ==
* Question: c("A", "B", NA) != "" ?
* [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'>
Answer: TRUE TRUE NA
# 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
</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
* Question: which(c("A", "B", NA) != "") ?
# docker run --net=host -it --rm r-base:3.6.0
Answer: 1 2
> 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>
* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
Answer: TRUE TRUE FALSE


== Regular Expression ==
* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
See [[Regular_expression|here]].
Answer: TRUE TRUE FALSE


== Read rrd file ==
'''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.
* 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/


== file, connection ==
Don't just use x != "" OR !is.na(x).
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
* read() and write()
* read.table() and write.table()
<syntaxhighlight lang='rsplus'>
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)
</syntaxhighlight>


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


== read/manipulate binary data ==
== Constant and 'L' ==
* x <- readBin(fn, raw(), file.info(fn)$size)
Add 'L' after a constant. For example,
* rawToChar(x[1:16])
{{Pre}}
* See Biostrings C API
for(i in 1L:n) { }


== String Manipulation ==
if (max.lines > 0L) { }
* [http://gastonsanchez.com/blog/resources/how-to/2013/09/22/Handling-and-Processing-Strings-in-R.html ebook] by Gaston Sanchez.
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]


== HTTPs connection ==
label <- paste0(n-i+1L, ": ")
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)
n <- length(x)if(n == 0L) { }
 
== setInternet2 ==
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.
 
Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
<pre>
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)
</pre>
</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.
== Vector/Arrays ==
<pre>
R indexes arrays from 1 like Fortran, not from 0 like C or Python.
R.Version()$"svn rev"
</pre>


If setInternet2(T), then https protocol is supported in download.file().
=== remove integer(0) ===
[https://stackoverflow.com/a/27980810 How to remove integer(0) from a vector?]


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.
=== Append some elements ===
[https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ append() and its after argument]


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].
=== setNames() ===
Assign names to a vector


'''R up to 3.2.2'''
<pre>
<pre>
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
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>
</pre>
See also
* <src/include/Internal.h> (declare do_setInternet2()),
* <src/main/names.c> (show do_setInternet2() in C)
* <src/main/internet.c>  (define do_setInternet2() in C).


Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file.  If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2, it is also possible to download from https by setIneternet2(T).
== Factor ==
=== labels argument ===
We can specify the factor levels and new labels using the factor() function.
 
{{Pre}}
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))


'''R 3.3.0'''
factor(rev(letters[1:3]), labels = c("A", "B", "C"))
<pre>
# C B A
setInternet2 <- function(use = TRUE) {
# Levels: A B C
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>
</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.
=== 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>
* 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


== File operation ==
library(tidyverse); library(magrittr)
* list.files()
set.seed(1)
* file.info()
breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
* dir.create()
x <- runif(50)
* file.create()
bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
* file.copy()


== read/download/source a file from internet ==
data.frame(sc=x, bins=bins) %>%
=== Simple text file http ===
  group_by(bins) %>%
  summarise(n=n()) %>%
  ggplot(aes(x = bins, y = n)) +
    geom_col(color = "black", fill = "#90AACB") +
    theme_minimal() +
    theme(axis.text.x = element_text(angle = 90)) +
    theme(legend.position = "none") + coord_flip()
</pre>
<li>[https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html A Guide to Using the cut() Function in R]
<li>[https://youtu.be/7oyiPBjLAWY?t=2480 tibble object]
{{Pre}}
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
      age_cat = case_when(
          age_yrs < 2 ~ "baby",
          age_yrs < 13 ~ "kid",
          age_yrs < 20 ~ "teen",
          TRUE        ~ "adult")
)
</pre>
</li>
<li>[https://youtu.be/JsNqXLl3eFc?t=96 R tip: Learn dplyr’s case_when() function]
<pre>
<pre>
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
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>
</pre>
<li>
</ul>


=== Zip file and url() function ===
=== How to change one of the level to NA ===
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
<pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
x <- factor(c("a", "b", "c", "NotPerformed"))
source(con)
levels(x)[levels(x) == 'NotPerformed'] <- NA
close(con)
</pre>
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.


Another example of using url() is
[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>
<pre>
load(url("http:/www.example.com/example.RData"))
unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
</pre>
</pre>


=== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ===
=== droplevels() ===
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.
[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.


=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
=== 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>
<pre>
require(RCurl)
# Syntax:
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
# newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
read.csv(textConnection(myCsv))
 
(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>
</pre>
<li>[http://www.deeplytrivial.com/2020/05/statistics-sunday-my-2019-reading.html Statistics Sunday: My 2019 Reading] (reorder function)
</ul>


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


=== Github files https using RCurl package ===
ordered(levels=c("a", "b", "c"))
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
# Error in factor(x, ..., ordered = TRUE) :
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
#  argument "x" is missing, with no default
<pre>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",  
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
</pre>
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package


== summarytools: create summary tables for vectors and data frames ==
== Data frame ==
https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and 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/
 
=== 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.


== Create publication tables using '''tables''' package ==
In R 4.0.0, [https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/ stringAsFactors=FALSE] will be default. This also affects read.table() function.
See 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,
=== check.names = FALSE ===
<syntaxhighlight lang='rsplus'>
Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.
> library(tables)
<pre>
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
  1a 2a
                                                 
1 1 1
                Sepal.Length      Sepal.Width   
2 2  2
  Species    n  mean        sd  mean        sd  
> data.frame("1a"=1:2, "2a"=1:2) # default
  setosa      50 5.01        0.35 3.43        0.38
  X1a X2a
versicolor  50 5.94        0.52 2.77        0.31
1   1   1
  virginica  50 6.59        0.64 2.97        0.32
2   2   2
All        150 5.84        0.83 3.06        0.44
> str(iris)
'data.frame':  150 obs. of  5 variables:
$ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
$ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
$ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
$ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
$ Species    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
</pre>
</pre>
and
 
=== Create unique rownames: make.unique() ===
<pre>
<pre>
# This example shows some of the less common options       
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
rownames(mydf) <- make.unique(groupCodes)
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
</pre>
> 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)
</syntaxhighlight>


== ClinReport: Statistical Reporting in Clinical Trials ==
=== data.frame() will change rownames ===
https://cran.r-project.org/web/packages/ClinReport/index.html
<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>


== Append figures to PDF files ==
=== Print a data frame without rownames ===
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
<pre>
# Method 1.  
rownames(df1) <- NULL


== Extracting tables from PDFs ==  
# Method 2.
* [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'''.
print(df1, row.names = FALSE)
* [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'>
</pre>
library(pdftools)
 
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
=== Convert data frame factor columns to characters ===
txt <- pdf_text(pdf_file) # length = number of pages
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
# Suppose the table we are interested in is on page 1
{{Pre}}
cat(txt[1]) # Good but not in a data frame format
# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
 
# Method 2:
bob[] <- lapply(bob, as.character)
</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)


pdf_data(pdf_file)[[1]]  # data frame/tibble format
=== data.frame to vector ===
</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'>
df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))
sudo apt install poppler-utils
pdftotext -layout input.pdf output.txt
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</syntaxhighlight>


== Create flat tables in R console using ftable() ==
class(df)
<syntaxhighlight lang='rsplus'>
# [1] "data.frame"
> ftable(Titanic, row.vars = 1:3)
class(t(df))
                  Survived  No Yes
# [1] "matrix" "array"
Class Sex    Age                 
class(unlist(df))
1st  Male  Child            0   5
# [1] "numeric"
            Adult          118  57
 
      Female Child            0  1
# Method 1: Convert data frame to matrix using as.matrix()
            Adult            4 140
# and then Convert matrix to vector using as.vector() or c()
2nd  Male  Child            0  11
mat <- as.matrix(df)
            Adult          154  14
vec1 <- as.vector(mat)   # [1] 1 2 3 4 5 6
      Female Child            0  13
vec2 <- c(mat)
            Adult          13  80
 
3rd  Male  Child          35  13
# Method 2: Convert data frame to matrix using t()/transpose
            Adult          387  75
# and then Convert matrix to vector using as.vector() or c()
      Female Child          17  14
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
            Adult          89  76
vec4 <- c(t(df))
Crew  Male  Child            0  0
 
            Adult          670 192
# Not working
      Female Child            0  0
as.vector(df)
            Adult            3 20
# $x
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
# [1] 1 2 3
            Survived  No Yes
# $y
Class Sex                   
# [1] 4 5 6
1st  Male            118 62
 
      Female            4 141
# Method 3: unlist() - easiest solution
2nd  Male            154 25
unlist(df)
      Female          13 93
# x1 x2 x3 y1 y2 y3
3rd  Male            422 88
# 1 2 3 4 5 6
      Female          106 90
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
Crew Male            670 192
# [1] 1 2 3 4 5 6
      Female            3  20
</pre>
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
Q: Why as.vector(df) cannot convert a data frame into a vector?
            Survived  No Yes
 
Sex    Class               
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.
Male  1st            118  62
 
      2nd            154  25
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.
      3rd            422  88
 
      Crew          670 192
=== Using cbind() to merge vectors together? ===
Female 1st              4 141
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.
      2nd            13  93
 
      3rd            106  90
=== cbind NULL and data.frame ===
      Crew            3  20
[https://9to5tutorial.com/cbind-can-t-combine-null-with-dataframe cbind can't combine NULL with dataframe]. Add as.matrix() will fix the problem.
> str(Titanic)
 
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
=== merge ===
- attr(*, "dimnames")=List of 4
* [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].
  ..$ Class  : chr [1:4] "1st" "2nd" "3rd" "Crew"
* [https://www.geeksforgeeks.org/merge-dataframes-by-row-names-in-r/ Merge DataFrames by Row Names in R]
  ..$ Sex    : chr [1:2] "Male" "Female"
* [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]
  ..$ Age    : chr [1:2] "Child" "Adult"
* [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]
  ..$ Survived: chr [1:2] "No" "Yes"
 
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
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
> x
<pre>
          gear  3  4  5
class(df1); class(df2)
cyl vs am             
# [1] "data.frame" # 2 x 2
4   0  0       0  0  0
# [1] "matrix" "array" # 52439 x 2
      1        0 0 1
rownames(df1)
    1  0       1  2  0
# [1] "A1CF"     "A1BG-AS1"
      1        0 6  1
merge(df1, df2[c(9109, 44999), ], by=0)
6   0 0        0  0  0
#   Row.names 786-0 A498 ACH-000001 ACH-000002
      1       0 1
# 1 A1BG-AS1    0   0   7.321358  6.908333
    0        2 2  0
# 2      A1CF    0   0   3.011470  1.189578
      1        0  0  0
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
8   0 0      12  0  0
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
      1       0 0 2
#   Row.names 786-0 A498 ACH-000001 ACH-000002
     0       0  0  0
# 1     A1CF    0   0    3.01147  1.189578
      1       0  0  0
rownames(df2)[38958:38959]
> ftable(x, row.vars = c(2, 4))
# [1] "ITFG2-AS1" "ADGRD1-AS1"
        cyl  4    6    8 
 
        am  0  1  0  1  0  1
rownames(df1)[2] <- "A1BGAS1"
vs gear                     
rownames(df2)[44999] <- "A1BGAS1"
0  3        0  0  0  0 12  0
merge(df1, df2, by= 0)
  4        0  0  0  2  0  0
#   Row.names 786-0 A498 ACH-000001 ACH-000002
  5        0  1  0  1  0  2
# 1   A1BGAS1    0   0   7.321358  6.908333
3         1  0  2  0  0  0
# 2     A1CF     0   0   3.011470  1.189578
  4        2  6 2 0  0  0
</pre>
  5        0  1  0  0  0  0
 
>  
=== is.matrix: data.frame is not necessarily a matrix ===
> ## Start with expressions, use table()'s "dnn" to change labels
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.
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
 
        dnn = c("Cylinders", "V/S", "Transmission", "Gears"))
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
 
=== 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.
* [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.
 
<syntaxhighlight lang='r'>
# 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>


          Cylinders    4    6    8 
* 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.
          Transmission  0  1 1
* 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).
V/S Gears                             
* See the following example where as.matrix() and data.matrix() return different resuls.
3                   0  0  0  0 12  0
<syntaxhighlight lang='r'>
    4                  0  0  0  2  0  0
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
    5                  0  1 1 2
mat <- as.matrix(df)
1   3                  1 0  2  0  0  0
mat
    4                  2  6  2  0  0  0
#      a  b  
    5                  0  1 0  0  0  0
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array"
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array"
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"
</syntaxhighlight>
</syntaxhighlight>


=== [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins] ===
=== matrix vs data.frame ===
Puts Arbitrary Margins On Multidimensional Tables Or Arrays
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
 
== tracemem, data type, copy ==
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]
 
== Tell if the current R is running in 32-bit or 64-bit mode ==
<pre>
<pre>
8 * .Machine$sizeof.pointer
Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
[1] "time"  "status" "treat"  "AKT1"  "BRAF"  "FLOT2"  "MTOR"  "PCK2"  "PIK3CA"
[10] "RAF1" 
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10
</pre>
</pre>
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 ==
Case 2:
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
{{Pre}}
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
* 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.
unique(ip1$Priority)
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])   # OK


== Handling length 2^31 and more in R 3.0.0 ==
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
 
unique(ip2$Priority)     # OK
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
<pre>
> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector
</pre>
</pre>


However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
The length of a matrix and a data frame is different.
<pre>
{{Pre}}
> system.time(x <- seq(1,2^31))
> length(matrix(1:6, 3, 2))
  user  system elapsed
[1] 6
  8.604  11.060 120.815
> length(data.frame(matrix(1:6, 3, 2)))
> length(x)
[1] 2
[1] 2147483648
> x[1]
> length(x)/2^20
   X1
[1] 2048
1  1
> gc()
2 2
            used    (Mb) gc trigger    (Mb)   max used    (Mb)
3
Ncells    183823    9.9    407500    21.8    350000    18.7
4  4
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
5  5
>
6  6
> x[[1]]
[1] 1 2 3 4 5 6
</pre>
</pre>
Note:
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.
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].


== NA in index ==
=== How to Remove Duplicates ===
* Question: what is seq(1, 3)[c(1, 2, NA)]?
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]


Answer: It will reserve the element with NA in indexing and return the value NA for it.
=== Convert a matrix (not data frame) of characters to numeric ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
{{Pre}}
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
    a    b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)    # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"


* Question: What is TRUE & NA?
> mode(tmp) <- "numeric"
Answer: NA
> sum(tmp)
[1] 1.917
</pre>


* Question: What is FALSE & NA?
=== Convert Data Frame Row to Vector ===
Answer: FALSE
as.numeric() or '''c()'''


* Question: c("A", "B", NA) != "" ?
=== Convert characters to integers ===
Answer: TRUE TRUE NA
mode(x) <- "integer"


* Question: which(c("A", "B", NA) != "") ?
=== Non-Standard Evaluation ===
Answer: 1 2
[https://thomasadventure.blog/posts/understanding-nse-part1/ Understanding Non-Standard Evaluation. Part 1: The Basics]


* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
=== Select Data Frame Columns in R ===
Answer: TRUE TRUE FALSE
This is part of series of DATA MANIPULATION IN R from [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ datanovia.com]


* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
Answer: TRUE TRUE FALSE
* 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


'''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.
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


Don't just use x != "" OR !is.na(x).
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>


== Constant and 'L' ==
=== Add columns to a data frame ===
Add 'L' after a constant. For example,
[https://datasciencetut.com/how-to-add-columns-to-a-data-frame-in-r/ How to add columns to a data frame in R]
<syntaxhighlight lang='rsplus'>
for(i in 1L:n) { }


if (max.lines > 0L) { }
=== 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) )


label <- paste0(n-i+1L, ": ")
# method 2
drop <- c("x","z")
df = mydata[,!(names(mydata) %in% drop)]


n <- length(x);  if(n == 0L) { }
# method 3: dplyr
</syntaxhighlight>
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>


== Factor ==
=== Remove Rows from the data frame ===
=== labels argument ===
[https://datasciencetut.com/remove-rows-from-the-data-frame-in-r/ Remove Rows from the data frame in R]
We can specify the factor levels and new labels using the factor() function.


<syntaxhighlight lang='rsplus'>
=== Danger of selecting rows from a data frame ===
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
<pre>
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
> dim(cars)
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))
[1] 50  2
</syntaxhighlight>
> data.frame(a=cars[1,], b=cars[2, ])
  a.speed a.dist b.speed b.dist
1      4      2      4    10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a  b
speed 4  4
dist  2 10
</pre>


=== Create a factor from a continuous variable: cut() ===
=== Creating data frame using structure() function ===
<syntaxhighlight lang='rsplus'>
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]
facVar <- cut(contVar, c(-3, 0, 2, 4), labels = c("low", "medium", "high"))
</syntaxhighlight>
Or [https://youtu.be/7oyiPBjLAWY?t=2480 a tibble object]
<syntaxhighlight lang='rsplus'>
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
      age_cat = case_when(
          age_yrs < 2 ~ "baby",
          age_yrs < 13 ~ "kid",
          age_yrs < 20 ~ "teen",
          TRUE        ~ "adult")
)
</syntaxhighlight>


== Data frame ==
=== Create an empty data.frame ===
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/
https://stackoverflow.com/questions/10689055/create-an-empty-data-frame
<pre>
# the column types default as logical per vector(), but are then overridden
a = data.frame(matrix(vector(), 5, 3,
              dimnames=list(c(), c("Date", "File", "User"))),
              stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)


=== stringsAsFactors = FALSE ===
# similar to above
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))


=== Convert data frame factor columns to characters ===
# different data type
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
a <- data.frame(x1 = character(),
<syntaxhighlight lang='rsplus'>
                x2 = numeric(),
# Method 1:
                x3 = factor(),
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
                stringsAsFactors = FALSE)
</pre>


# Method 2:
=== Objects from subsetting a row in a data frame vs matrix ===
bob[] <- lapply(bob, as.character)
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
</syntaxhighlight>
<ul>
 
<li>Subsetting creates repeated rows. This will create unexpected rownames.
=== data.frame to vector ===
<pre>
<syntaxhighlight lang='rsplus'>
R> z <- data.frame(x=1:3, y=2:4)
> a= matrix(1:6, 2,3)
R> rownames(z) <- letters[1:3]
> rownames(a) <- c("a", "b")
R> rownames(z)[c(1,1)]
> colnames(a) <- c("x", "y", "z")
[1] "a" "a"
> a
R> rownames(z[c(1,1),])
  x y z
[1] "a"   "a.1"
a 1 3 5
R> z[c(1,1), ]
b 2 4 6
    x y
> unlist(data.frame(a))
a   1 2
x1 x2 y1 y2 z1 z2
a.1 1 2
  1  2 4 5 6
</pre>
</syntaxhighlight>
</li>
<li>[https://stackoverflow.com/a/2545548 Convert a dataframe to a vector (by rows)] The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
{{Pre}}
str(trainData)
# 'data.frame': 503 obs. of  500 variables:
#  $ bm001: num 0.429 1 -0.5 1.415 -1.899 ...
# $ bm002: num 0.0568 1 0.5 0.3556 -1.16 ...
# ...
trainData[1:3, 1:3]
#        bm001      bm002    bm003
# 1 0.4289449 0.05676296 1.657966
# 2 1.0000000 1.00000000 1.000000
# 3 -0.5000000 0.50000000 0.500000
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
# Warning message:
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, :
#  row names were found from a short variable and have been discarded
</pre>
 
'trees' data from the 'datasets' package
<pre>
trees[1:3,]
#  Girth Height Volume
# 1  8.3    70  10.3
# 2  8.6    65  10.3
# 3  8.8    63  10.2


=== merge ===
# Wrong ways:
[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]
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


=== matrix vs data.frame ===
# Right ways:
<syntaxhighlight lang='rsplus'>
# method 1: dropping row names
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,])))
unique(ip1$Priority)
# OR
# Error in ip1$Priority : $ operator is invalid for atomic vectors
data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
unique(ip1[, "Priority"])  # OK
#  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


ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
# Method 3: convert a data frame to a matrix
unique(ip2$Priority)    # OK
is.matrix(trees)
</syntaxhighlight>
# [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


The length of a matrix and a data frame is different.
dim(trees[1,])
<syntaxhighlight lang='rsplus'>
# [1] 1 3
> length(matrix(1:6, 3, 2))
dim(trees2[1, ])
[1] 6
# NULL
> length(data.frame(matrix(1:6, 3, 2)))
trees[1, ] # notice the row name '1' on the left hand side
[1] 2
#   Girth Height Volume
> x[1]
# 1   8.3    70  10.3
   X1
trees2[1, ]
1 1
# Girth Height Volume
2 2
#    8.3   70.0  10.3
3 3
</pre>
4  4
</li>
5  5
</ul>
6  6
> x[[1]]
[1] 1 2 3 4 5 6
</syntaxhighlight>
So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.


=== Convert a matrix (not data frame) of characters to numeric ===
=== Convert a list to data frame ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
[https://www.statology.org/convert-list-to-data-frame-r/ How to Convert a List to a Data Frame in R].
<syntaxhighlight lang='rsplus'>
<pre>
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
# method 1
    a    b
data.frame(t(sapply(my_list,c)))
1 0.12 0.567
 
2 0.34 0.890
# method 2
> is.data.frame(tmp) # FALSE
library(dplyr)
> is.matrix(tmp)     # TRUE
bind_rows(my_list) # OR bind_cols(my_list)
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"


> mode(tmp) <- "numeric"
# method 3
> sum(tmp)
library(data.table)
[1] 1.917
rbindlist(my_list)
</syntaxhighlight>
</pre>


=== Select Data Frame Columns in R ===
=== tibble and data.table ===
This is part of series of DATA MANIPULATION IN R from [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ datanovia.com]
* [[R#tibble | tibble]]
* [[Tidyverse#data.table|data.table]]


* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
=== Clean  a dataset ===
* select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
[https://finnstats.com/index.php/2021/04/04/how-to-clean-the-datasets-in-r/ How to clean the datasets in R]
* select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
* Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names


=== Creating data frame using structure() function ===
== matrix ==
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]


=== Warning: row names were found from a short variable and have been discarded ===
=== Define and subset a matrix ===
https://stackoverflow.com/a/23534617
* [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]'''.


My example:
<pre>
<syntaxhighlight lang='rsplus'>
data <- c(2, 4, 7, 5, 10, 1)
str(trainData)
A <- matrix(data, ncol = 3)
# 'data.frame': 503 obs. of  500 variables:
print(A)
#  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
#      [,1] [,2] [,3]
#  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
# [1,]   2    7  10
# ...
# [2,]   4    5    1
trainData[1:3, 1:3]
#       bm001     bm002    bm003
# 1  0.4289449 0.05676296 1.657966
# 2  1.0000000 1.00000000 1.000000
# 3 -0.5000000 0.50000000 0.500000
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
# Warning message:
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
#  row names were found from a short variable and have been discarded
</syntaxhighlight>


== matrix (column-major order) multiply a vector ==
A[1:1, 2:3, drop=F]
* [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) is following the column-major order]
#      [,1] [,2]
# [1,]   7  10
</pre>


<syntaxhighlight lang='rsplus'>
=== Prevent automatic conversion of single column to vector ===
> matrix(1:6, 3,2)
use '''drop = FALSE''' such as mat[, 1, drop = FALSE].
 
=== complete.cases(): remove rows with missing in any column ===
It works on a sequence of vectors, matrices and data frames.
 
=== NROW vs nrow ===
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/nrow ?nrow]. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.
 
=== matrix (column-major order) multiply a vector ===
* Matrices in R [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) are stored in a column-major order]. It means array slice A[,1] are contiguous.
 
{{Pre}}
> matrix(1:6, 3,2)
     [,1] [,2]
     [,1] [,2]
[1,]    1    4
[1,]    1    4
[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 4,648: 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 4,654: 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'>
> help.start()
If the browser launched by 'xdg-open' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
> /usr/bin/xdg-open: 461: /usr/bin/xdg-open: x-www-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: firefox: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: mozilla: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: epiphany: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: konqueror: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: chromium-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: google-chrome: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links2: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: lynx: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: w3m: not found
xdg-open: no method available for opening 'http://127.0.0.1:27919/doc/html/index.html'
</syntaxhighlight>
 
The solution is to put
<pre>
<pre>
options(browser='seamonkey')
options(browser='seamonkey')
Line 4,689: Line 5,202:


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 4,702: 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 4,727: 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 4,753: 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 4,768: 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 ===
[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 ===
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1
* 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 4,787: Line 5,367:
paste("name", formatC(a, width=3, flag="0"), sep="_")
paste("name", formatC(a, width=3, flag="0"), sep="_")
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
</syntaxhighlight>


=== sprintf does not print ===
# gsub()
Use cat() or print() outside sprintf(). sprintf() do not print in a non interactive mode.
paste0("bm", gsub(" ", "0", format(5:15)))
<syntaxhighlight lang='rsplus'>
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
cat(sprintf('%5.2f\t%i\n',1.234, l234))  
</pre>
</syntaxhighlight>
 
=== 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>
 
=== Format(x, scientific = TRUE) ===
Print numeric data in exponential format, so .0001 prints as 1e-4


== Creating publication quality graphs in R ==
== Creating publication quality graphs in R ==
Line 4,805: 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 4,863: 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 4,882: 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 4,891: 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 4,904: 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 4,911: 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 4,924: Line 5,508:
})
})
</pre>
</pre>
[[File:RugFunction.png|200px]]
[[:File:RugFunction.png]]


See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.
See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.


== Identify/Locate Points in a Scatter Plot ==
== 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 4,936: Line 5,531:
== Draw Color Palette ==
== Draw Color Palette ==
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
=== Default palette before R 4.0 ===
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 4,942: 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 4,962: 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 4,979: 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 5,102: Line 5,787:
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]


== as.formula() ==
=== Careful on NA value ===
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<pre>
R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3  3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6
</pre>
 
=== Implicit looping ===
<pre>
set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]    # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7
</pre>
 
== modelling ==
=== update() ===
* [https://www.rdocumentation.org/packages/stats/versions/3.6.1/topics/update ?update]
* [https://stackoverflow.com/a/5118337 Reusing a Model Built in R]
 
=== Extract all variable names in lm(), glm(), ... ===
all.vars(formula(Model)[-2])
 
=== as.formula(): use a string in formula in lm(), glm(), ... ===
* [https://www.r-bloggers.com/2019/08/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
<syntaxhighlight lang='rsplus'>
{{Pre}}
? as.formula
? as.formula
xnam <- paste("x", 1:25, sep="")
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</syntaxhighlight>
</pre>
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
<syntaxhighlight lang='rsplus'>
{{Pre}}
outcome <- "mpg"
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")
variables <- c("cyl", "disp", "hp", "carb")
Line 5,158: Line 5,883:
(Intercept)          cyl        disp          hp        carb   
(Intercept)          cyl        disp          hp        carb   
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
</syntaxhighlight>
</pre>
* [https://statisticaloddsandends.wordpress.com/2019/08/24/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula] 1. as.formula() 2. subset by [[i]] 3. get() 4. eval(parse()).
 
=== reformulate ===
[https://www.r-bloggers.com/2023/06/simplifying-model-formulas-with-the-r-function-reformulate/ Simplifying Model Formulas with the R Function ‘reformulate()’]
 
=== I() function ===
I() means isolates. See [https://stackoverflow.com/a/24192745 What does the capital letter "I" in R linear regression formula mean?],  [https://stackoverflow.com/a/8055683 In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)]
 
=== Aggregating results from linear model ===
https://stats.stackexchange.com/a/6862
 
== Replacement function "fun(x) <- a" ==
[https://stackoverflow.com/questions/11563154/what-are-replacement-functions-in-r What are Replacement Functions in R?]
<pre>
R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
    x[x > value] <- Inf
    x
}
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]  1  3 Inf Inf
 
R> "cutoff<-"(x = xx, value = 65)
[1]  1  3 Inf Inf
</pre>
The statement '''fun(x) <- a''' and R will read '''x <- "fun<-"(x,a) '''


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


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


* '''Debug a S4 function'''
<syntaxhighlight lang='rsplus'>
> library(genefilter) # Bioconductor
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
> showMethods("nsFilter")
Line 5,188: Line 5,941:
eset="ExpressionSet"
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")
> debug(nsFilter, signature="ExpressionSet")
</syntaxhighlight>
 
library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code
</pre>
See the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L367 source code] of '''normalizationFactors<-''' (setReplaceMethod() is used) and the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L385 source code] of '''estimateSizeFactors()'''. We can see how ''avgTxLength'' was used in estimateNormFactors().
 
Another example
<pre>
library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)
 
showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"
 
debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz,
#      normalization = ssgsea.norm, verbose = verbose,
#      BPPARAM = BPPARAM))
 
isdebugged("gsva")
# [1] TRUE
undebug(gsva)
</pre>


* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
<syntaxhighlight lang='rsplus'>
{{Pre}}
library(IRanges)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir <- IRanges(start=c(10, 20, 30), width=5)
Line 5,222: 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 5,234: 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 5,249: Line 6,075:


For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
<syntaxhighlight lang='rsplus'>
{{Pre}}
> mcols(ddsNoPrior[genes, ])
> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
DataFrame with 2 rows and 21 columns
Line 5,268: Line 6,094:
1      TRUE        3  210.4045 0.2648753
1      TRUE        3  210.4045 0.2648753
2      TRUE        9  243.7455 0.3248949
2      TRUE        9  243.7455 0.3248949
</syntaxhighlight>
</pre>
 
== Pipe ==
* [https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
* [https://win-vector.com/2020/12/05/r-is-getting-an-official-pipe-operator/ R is Getting an Official Pipe Operator], [https://win-vector.com/2020/12/07/my-opinion-on-rs-upcoming-pipe/ My Opinion on R’s Upcoming Pipe]
<ul>
<li> a(b(x)) vs '''x |> b() |> a()'''. See [https://twitter.com/henrikbengtsson/status/1335328090390597632 this tweet] in R-dev 2020-12-04.
<pre>
e0 <- quote(a(b(x)))
e1 <- quote(x |> b() |> a())
identical(e0, e1)
</pre>
</li>
<li>
[https://selbydavid.com/2021/05/18/pipes/ There are now 3 different R pipes]
</li>
<li>[https://stackoverflow.com/a/67629310 Error: The pipe operator requires a function call as RHS].
<pre>
# native pipe
foo |> bar()
# magrittr pipe
foo %>% bar
</pre>
</li>
<li>[https://www.infoworld.com/article/3621369/use-the-new-r-pipe-built-into-r-41.html Use the new R pipe built into R 4.1] </li>
<li>[https://towardsdatascience.com/the-new-native-pipe-operator-in-r-cbc5fa8a37bd The New Native Pipe Operator in R] </li>
<li>[https://ivelasq.rbind.io/blog/understanding-the-r-pipe/ Understanding the native R pipe |> ] </li>
</ul>
 
Packages take advantage of pipes
<ul>
<li>[https://cran.r-project.org/web/packages/rstatix/index.html rstatix]: Pipe-Friendly Framework for Basic Statistical Tests
</ul>


== findInterval() ==
== findInterval() ==
Line 5,274: Line 6,132:
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
== Assign operator ==
* Earlier versions of R used underscore (_) as an assignment operator.
* [https://developer.r-project.org/equalAssign.html Assignments with the = Operator]
* In R 1.8.0 (2003), the assign operator has been removed. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
* In R 1.9.0 (2004), "_" is allowed in valid names. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
: [[File:R162.png|200px]]
== Operator precedence ==
The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.


== order(), rank() and sort() ==
== order(), rank() and sort() ==
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
<syntaxhighlight lang='rsplus'>
<pre>
> x = sample(10)
> x = sample(10)
> x
> x
Line 5,292: 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"
 
# linux
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.
 
The difference is related to locale. See
 
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/locales ?locales] in R
* On OS, type '''locale'''
* [https://stackoverflow.com/questions/39171613/sort-produces-different-results-in-ubuntu-and-windows sort() produces different results in Ubuntu and Windows]
* To fix the inconsistency problem, we can set the locale in R code to "C" or use the stringr package (the locale is part of [https://www.rdocumentation.org/packages/stringr/versions/1.4.0/topics/str_order str_order()]'s arguments).
<pre>
# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2
 
# 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 ==
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.


Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html. help].
[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
* Usage
: <syntaxhighlight lang='rsplus'>
{{Pre}}
do.call(what, args, quote = FALSE, envir = parent.frame())
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.
# what: either a function or a non-empty character string naming the function to be called.
Line 5,307: Line 6,227:
# 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 5,317: 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 5,347: 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.
{{Pre}}
> A <- 2
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A) 
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env) 
[1] 4
> do.call(f, list(A), envir = env) 
[1] 2                      # Why?
 
> 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
</pre>
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
{{Pre}}
> foo <- function(a=1, b=2, ...) {
        list(arg=do.call(c, as.list(match.call())[-1]))
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a
1
> foo(a=1, b=2, c=3)
$arg
a b c
1 2 3
</pre>
* do.call() + switch(). See [https://github.com/satijalab/seurat/blob/13b615c27eeeac85e5c928aa752197ac224339b9/R/preprocessing.R#L2450 an example] from Seurat::NormalizeData.
<pre>
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() 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 5,379: Line 6,375:
[11,] "a"  "3"  "-"  
[11,] "a"  "3"  "-"  
[12,] "b"  "3"  "-"  
[12,] "b"  "3"  "-"  
</syntaxhighlight>
</pre>
* ''environment'' and ''quote'' arguments.
* 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}}
> A <- 2
set.seed(1)
> f <- function(x) print(x^2)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
> env <- new.env()
# [1] 1.326108 9.885284
> assign("A", 10, envir = env)
set.seed(1)
> assign("f", f, envir = env)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
> f <- function(x) print(x)
dim(x) # [1]  2 1000
> f(A)  
rowMeans(x)
[1] 2
# [1] 1.032209 10.104131
> do.call("f", list(A))
</pre>
[1] 2
{{Pre}}
> do.call("f", list(A), envir = env)
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
[1] 4
# [1] 1.326108 9.885284
> do.call(f, list(A), envir = env)  
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))
[1] 2                      # Why?
</pre>
 
=== do.call vs lapply ===
[https://stackoverflow.com/a/10801883 What's the difference between lapply and do.call?] It seems to me the best usage is combining both functions: '''do.call(..., lapply())'''


> 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
</syntaxhighlight>
=== do.call vs lapply ===
https://stackoverflow.com/a/10801883
* 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 5,434: 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()]. 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 5,450: Line 6,437:
[[3]]
[[3]]
[1] 8
[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)
> do.call(rbind,X)
     [,1] [,2] [,3]
     [,1] [,2] [,3]
Line 5,471: 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].
{{Pre}}
> 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"
</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 5,513: 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 5,524: 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 5,548: 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 5,567: 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
* 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'>
mode(quote(1:10))
# [1] "call"
</syntaxhighlight>
* eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.  
** The '''parent.frame()''' is necessary in cases like the [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/update stats::update()] function used by [https://github.com/cran/glmnet/blob/master/R/relax.glmnet.R#L66 relax.glmnet()].
** Example:
::<syntaxhighlight lang='rsplus'>
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))
 
subset1 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR
 
subset2 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌
subset2(sample_df, a == condition) # 👍
</syntaxhighlight>
* deparse(expr) - turns unevaluated expressions into character strings. For example,  
* deparse(expr) - turns unevaluated expressions into character strings. For example,  
<pre>
:<syntaxhighlight lang='rsplus'>
> deparse(args(lm))
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "  
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "  
Line 5,597: 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 5,603: 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 5,625: 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?]


== Lazy evaluation in R functions arguments ==
== Functions ==
* http://adv-r.had.co.nz/Functions.html
* https://adv-r.hadley.nz/functions.html
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.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!


'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  
=== Function argument ===
[https://cran.r-project.org/doc/manuals/r-release/R-lang.html#Argument-matching Argument matching] from [https://cran.r-project.org/doc/manuals/r-release/R-lang.html R Language Definition] manual.
 
Argument matching is augmented by the functions
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.arg match.arg],
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.call match.call]
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.fun match.fun].
 
Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
 
=== Check function arguments ===
[https://blog.r-hub.io/2022/03/10/input-checking/ Checking the inputs of your R functions]: '''match.arg()''' , '''stopifnot()'''
 
'''stopifnot()''': function argument sanity check
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]. ''stopifnot'' is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.
<pre>
stopifnot(condition1, condition2, ...)
</pre>
</li>
<li>[https://rud.is/b/2020/05/19/mining-r-4-0-0-changelog-for-nuggets-of-gold-1-stopifnot/ Mining R 4.0.0 Changelog for Nuggets of Gold] </li>
</ul>
 
=== Lazy evaluation in R functions arguments ===
* http://adv-r.had.co.nz/Functions.html
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
* https://twitter.com/_wurli/status/1451459394009550850
 
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  


* Example 1. By default, R function arguments are lazy.
* Example 1. By default, R function arguments are lazy.
Line 5,681: Line 6,835:
}
}
</pre>
</pre>
=== Use of functions as arguments ===
[https://www.njtierney.com/post/2019/09/29/unexpected-function/ Just Quickly: The unexpected use of functions as arguments]
=== body() ===
[https://stackoverflow.com/a/51548945 Remove top axis title base plot]
=== Return functions in R ===
* [https://win-vector.com/2015/04/03/how-and-why-to-return-functions-in-r/ How and why to return functions in R]
* See the doc & example from [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/taskCallback taskCallback - Create an R-level task callback manager]. [https://developer.r-project.org/TaskHandlers.pdf Top-level Task Callbacks in R].
* [https://purrple.cat/blog/2017/05/28/turn-r-users-insane-with-evil/ Turn R users insane with evil]
=== anonymous function ===
In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a '''lambda function is defined without a name''', while a regular function is defined with a name.
<ul>
<li>See [[Tidyverse#Anonymous_functions|Tidyverse]] page
<li>But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, '''lapply(list(1,2,3), function(x) { x * x }) '''
<li>you can use lambda functions with many other functions in R that take a function as an argument. Some examples include '''sapply, apply, vapply, mapply, Map, Reduce, Filter''', and '''Find'''. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
<pre>
Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
</pre>
<li>[https://coolbutuseless.github.io/2019/03/13/anonymous-functions-in-r-part-1/ purrr anonymous function]
<li>[https://towardsdatascience.com/the-new-pipe-and-anonymous-function-syntax-in-r-54d98861014c The new pipe and anonymous function syntax in R 4.1.0]
<li>[http://adv-r.had.co.nz/Functional-programming.html#anonymous-functions Functional programming] from Advanced R
<li>[https://www.projectpro.io/recipes/what-are-anonymous-functions-r What are anonymous functions in R].
<syntaxhighlight lang='rsplus'>
> (function(x) x * x)(3)
[1] 9
> (\(x) x * x)(3)
[1] 9
</syntaxhighlight>
</ul>


== Backtick sign, infix/prefix/postfix operators ==  
== 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 5,692: 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 5,717: 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 5,738: 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)


== Using list type ==
== List data type ==
=== Avoid if-else or switch ===
=== Create an empty list ===
?plot.stepfun.
<pre>
out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)
 
outlist <- as.list(seq(nfolds))
</pre>
 
=== Using $ in R on a List ===
[https://r-lang.com/dollar-sign-in-r-with-example/ How to Use Dollar sign in R]
 
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5
</pre>
 
=== Descend recursively through lists ===
<nowiki>x[[c(5,3)]] </nowiki> is the same as <nowiki>x[[5]][[3]]</nowiki>. See [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Extract ?Extract].
 
=== Avoid if-else or switch ===
?plot.stepfun.
<pre>
<pre>
y0 <- c(1,2,4,3)
y0 <- c(1,2,4,3)
Line 5,767: Line 7,012:
par(op)
par(op)
</pre>
</pre>
[[File:StepfunExample.svg|400px]]
[[:File:StepfunExample.svg]]


== Open a new Window device ==
== Open a new Window device ==
Line 5,775: 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 5,801: 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).


=== pch ===
=== move axis title closer to axis ===
[[File:R pch.png|250px]]
* [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>


([https://www.statmethods.net/advgraphs/parameters.html figure source])
=== pch and point shapes ===
[[:File:R pch.png]]
 
See [https://www.statmethods.net/advgraphs/parameters.html here].


* 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|250px]]
[[:File:R lty.png]]


([http://www.sthda.com/english/wiki/line-types-in-r-lty figure source])
[https://finnstats.com/index.php/2021/06/11/line-types-in-r-lty-for-r-baseplot-and-ggplot/ Line types in R: Ultimate Guide For R Baseplot and ggplot]
 
See [http://www.sthda.com/english/wiki/line-types-in-r-lty here].
 
ggpubr::show_line_types()


=== las (label style) ===
=== las (label style) ===
Line 5,826: 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 5,845: 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 5,854: Line 7,167:
identical(mod, mod2, ignore.environment = TRUE)
identical(mod, mod2, ignore.environment = TRUE)
</pre>
</pre>
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/dput dput]: Writes an ASCII text representation of an R object. The object name is not written (unlike '''dump''').
{{Pre}}
$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato",
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos",
"ast", "trig", "platelet", "protime", "stage")
> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5,
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L,
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")
</pre>
=== User 'verbose = TRUE' in load() ===
When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.
=== What are RDS files anyways ===
[https://www.statworx.com/de/blog/archive-existing-rds-files/ Archive Existing RDS Files]


== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
* ==: exact match
* ==: exact match
* all.equal: compare R objects x and y testing ‘near equality’
* '''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.
* identical: The safe and reliable way to test two objects for being exactly equal.
<syntaxhighlight lang='rsplus'>
{{Pre}}
x <- 1.0; y <- 0.99999999999
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
all.equal(x, y)
Line 5,865: Line 7,200:
identical(x, y)
identical(x, y)
# [1] FALSE
# [1] FALSE
</syntaxhighlight>
</pre>


See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
Be careful about using "==" to return an index of matches in the case of data with missing values.
<pre>
R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1
</pre>
 
See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
 
I found a case when I compare two objects where 1 is generated in ''Linux'' and the other is generated in ''macOS'' that identical() gives FALSE but '''all.equal()''' returns TRUE. The difference has a magnitude only e-17.
 
=== waldo ===
* https://waldo.r-lib.org/ or [https://cloud.r-project.org/web/packages/waldo/index.html CRAN]. Find and concisely describe the difference between a pair of R objects.
* [https://predictivehacks.com/how-to-compare-objects-in-r/ How To Compare Objects In R]
 
=== diffobj: Compare/Diff R Objects ===
https://cran.r-project.org/web/packages/diffobj/index.html


== testhat ==
== 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 ==
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
[https://cran.r-project.org/web/packages/ttdo/index.html ttdo] adds support of the 'diffobj' package for 'diff'-style comparison of R objects.


== Numerical Pitfall ==
== 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 5,892: 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
** Comments
** Strings
** Loops
** Code Sharing
**Good Programming Practices


== How to debug an R code ==
== How to debug an R code ==
=== Using assign() in functions ===
[[Debug#R|Debug R]]
For example, insert the following line to your function
<pre>
assign(envir=globalenv(), "GlobalVar", localvar)
</pre>
 
== Debug lapply()/sapply() ==
* https://stackoverflow.com/questions/1395622/debugging-lapply-sapply-calls
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/recover.html. Use options(error=NULL) to turn it off.
 
== Debugging with RStudio ==
* https://www.rstudio.com/resources/videos/debugging-techniques-in-rstudio/
* https://github.com/ajmcoqui/debuggingRStudio/blob/master/RStudio_Debugging_Cheatsheet.pdf
* https://support.rstudio.com/hc/en-us/articles/205612627-Debugging-with-RStudio
 
== Debug R source code ==
=== Build R with debug information ===
* [[R#Build_R_from_its_source|R -> Build R from its source on Windows]]
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/ (defunct)
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/gdb.shtml (defunct)
* Build R with debug information (see the discussion [https://stackoverflow.com/a/30001096 here]). Cf [https://github.com/arraytools/r-build-output output messages] from running ./configure and make using the default options.
: <syntaxhighlight lang='bash'>
$ ./configure --help
$ ./configure --enable-R-shlib --with-valgrind-instrumentation=2 \
                              --with-system-valgrind-headers \
              CFLAGS='-g -O0 -fPIC' \
              FFLAGS='-g -O0 -fPIC' \
              CXXFLAGS='-g -O0 -fPIC' \
              FCFLAGS='-g -O0 -fPIC'
$ make -j4
$ sudo make install
</syntaxhighlight>
* [https://github.com/arraytools/r-debug My note of debugging cor() function]
* [https://vimeo.com/11937905 Using gdb to debug R packages with native code] (Video) The steps to debug is given below.
: <syntaxhighlight lang='bash'>
# Make sure to create a file <src/Makevars> with something like: CFLAGS=-ggdb -O0
# Or more generally
# CFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# CXXFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# FFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
 
$ tree nidemo
$ R CMD INSTALL nidemo
$ cat bug.R
$ R -f bug.R
$ R -d gdb
(gdb) r
> library(nidemo)
> Ctrl+C
(gdb) b nid_buggy_freq
(gdb) c  # continue
> buggy_freq("nidemo/DESCRIPTION") # stop at breakpoint 1
(gdb) list
(gdb) n # step through
(gdb) # press RETURN a few times until you see the bug
(gdb) d 1 # delete the first break point
(gdb) b Rf_error # R's C entry point for the error function
(gdb) c
> buggy_freq("nidemo/DESCRIPTION")
(gdb) bt 5 # last 5 stack frames
(gdb) frame 2
(gdb) list
(gdb) p freq_data
(gdb) p ans
(gdb) call Rf_PrintValues(ans)
(gdb) call Rf_PrintValues(fname)
(gdb) q
# Edit buggy.c
 
$ R CMD INSTALL nidemo # re-install the package
$ R -f bug.R
$ R -d gdb
(gdb) run
> source("bug.R") # error happened
(gdb) bt 5 # show the last 5 frames
(gdb) frame 2
(gdb) list
(gdb) frame 1
(gdb) list
(gdb) p file
(gdb) p fh
(gdb) q
# Edit buggy.c
 
$ R CMD INSTALL nidemo
$ R -f bug.R
</syntaxhighlight>
* [http://r-pkgs.had.co.nz/src.html Compiled code] from "R packages" by Hadley Wickham
* [https://www.bioconductor.org/developers/how-to/c-debugging/ Debugging C/C++ code] from Bioconductor (case study)
* Same idea for the Rcpp situation. See [https://stackoverflow.com/questions/21226337/what-are-productive-ways-to-debug-rcpp-compiled-code-loaded-in-r-on-os-x-maveri What are productive ways to debug Rcpp compiled code loaded in R (on OS X Mavericks)?]
 
=== .Call ===
* [https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Calling-_002eCall Writing R Extensions] manual.
* [http://adv-r.had.co.nz/C-interface.html R’s C interface] from Advanced R by Hadley Wickham
 
=== Registering native routines ===
https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Registering-native-routines
 
Pay attention to the prefix argument '''.fixes''' (eg .fixes = "C_") in '''useDynLib()''' function in the NAMESPACE file.
 
=== Example of debugging cor() function ===
Note that R's cor() function called a C function cor().
<pre>
stats::cor
....
.Call(C_cor, x, y, na.method, method == "kendall")
</pre>
 
A step-by-step screenshot of debugging using the GNU debugger '''gdb''' can be found on my Github repository https://github.com/arraytools/r-debug.


== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
Line 6,025: 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 6,039: 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 6,048: 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 6,066: 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 6,079: 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 6,092: 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]
m0 <-1
* https://rosettacode.org/wiki/Run-length_encoding
n <- 500
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
prob.tt <- .5
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package
p <- 500
 
b.corr.by <- 25
== citation() ==
b.corr <- .7
{{Pre}}
data <- data.frame(treat = rbinom(n, 1, prob.tt) - 0.5)
citation()
n.blocks <- p%/%b.corr.by
citation("MASS")
covMat <- diag(n.blocks) %x%
toBibtex(citation())
  matrix(b.corr^abs(matrix(1:b.corr.by, b.corr.by, b.corr.by, byrow = TRUE) -  
</pre>
                    matrix(1:b.corr.by, b.corr.by, b.corr.by)), b.corr.by, b.corr.by)
 
diag(covMat) <- 1
== R not responding request to interrupt stop process ==
data <- cbind(data, mvrnorm(n, rep(0, p), Sigma = covMat))
[https://stackoverflow.com/a/43172530 R not responding request to interrupt stop process]. ''R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts.'' It seems to match with the case I'm running (''dist()'' function).
range(data)
 
# Mac: -4.963827  4.133723
== Monitor memory usage ==
# Linux/Windows: -4.327635  4.408097
* x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
packageVersion("MASS")
* Windows: memory.size(max=TRUE)
# Mac: [1] ‘7.3.49’
* Linux
# Linux: [1] ‘7.3.49’
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.
# Windows: [1] ‘7.3.47’
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
 
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
R.version$version.string
* '''mem_used()''' function from [https://cran.r-project.org/web/packages/pryr/index.html pryr] package. It is not correct or useful if I use it to check the value compared to the memory returned by '''jobload''' in biowulf. So I cannot use it to see the memory used in running mclapply().  
# Mac: [1] "R version 3.4.3 (2017-11-30)"
* [https://cran.r-project.org/web/packages/peakRAM/index.html peakRAM]: Monitor the Total and Peak RAM Used by an Expression or Function
# Linux: [1] "R version 3.4.4 (2018-03-15)"
* [https://www.zxzyl.com/archives/1456/ Error: protect () : protection stack overflow] and [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Memory ?Memory]
# Windows: [1] "R version 3.4.3 (2017-11-30)"
 
References:
* [https://unix.stackexchange.com/questions/554/how-to-monitor-cpu-memory-usage-of-a-single-process How to monitor CPU/memory usage of a single process?]. ''htop -p $PID'' is recommended. It only shows the percentage of memory usage.
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
* [https://serverfault.com/a/264856 How can I see the memory usage of a Linux process?] ''pmap $PID | tail -n 1'' is recommended. It shows the memory usage in absolute value (eg 1722376K).
* [https://stackoverflow.com/a/6457769 How to check the amount of RAM in R] '''memfree <- as.numeric(system("awk '/MemFree/ {print $2}' /proc/meminfo", intern=TRUE)); memfree '''
 
== Monitor Data ==
[https://www.jstatsoft.org/article/view/v098i01?s=09 Monitoring Data in R with the lumberjack Package]
 
== Pushover ==
[https://rud.is/b/2020/01/29/monitoring-website-ssl-tls-certificate-expiration-times-with-r-openssl-pushoverr-and-dt/ Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}]
 
[https://cran.r-project.org/web/packages/pushoverr/ pushoverr]
 
= Resource =
== Books ==
* [https://forwards.github.io/rdevguide/ R Development Guide] R Contribution Working Group
* [https://rviews.rstudio.com/2021/11/04/bookdown-org/ An R Community Public Library] 2011-11-04
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
* [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]
<ul>
<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.
{{Pre}}
# 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
</pre>
</li>
</ul>
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro
* [https://rstats.wtf/ What They Forgot to Teach You About R] Jennifer Bryan & Jim Hester
* [http://knosof.co.uk/ESEUR/ Evidence-based Software Engineering] by Derek M. Jones
* [https://www.bigbookofr.com/index.html Big Book of R]
* [https://epirhandbook.com/?s=09 R for applied epidemiology and public health]
* [http://bendixcarstensen.com/EwR/ Epidemiology with R] and the [https://cran.r-project.org/web/packages/Epi/ Epi] package. [https://rdrr.io/cran/Epi/man/ci.lin.html ci.lin()] function to return the CI from glm() fit.
* [https://education.rstudio.com/learn/ RStudio &rarr; Finding Your Way To R]. Beginners/Intermediates/Experts
* [https://deepr.gagolewski.com/index.html Deep R Programming]
 
== Videos ==
[https://www.infoworld.com/article/3411819/do-more-with-r-video-tutorials.html “Do More with R” video tutorials]. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.
 
=== Webinar ===
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
 
== useR! ==
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]
* [http://www.user2019.fr/ UseR! 2019], [https://github.com/sowla/useR2019-materials tutorial], [https://www.mango-solutions.com/blog/user2019-roundup-workflow-reproducibility-and-friends Better workflow]
* [https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/playlists UseR! 2020 & 2021]
* [https://rviews.rstudio.com/2021/09/09/a-guide-to-binge-watching-r-medicine/ A Guide to Binge Watching R / Medicine 2021]
* [https://t.co/QBZwNoPJsC UseR! 2022]
 
== R consortium ==
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
 
== Blogs, Tips, Socials, Communities ==
* Google: revolutionanalytics In case you missed it
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]
* The R Blog <s>https://developer.r-project.org/Blog/public/</s> https://blog.r-project.org/
* [https://www.dataquest.io/blog/top-tips-for-learning-r-from-africa-rs-shelmith-kariuki/ Top Tips for Learning R from Africa R’s Shelmith Kariuki]
* [https://smach.github.io/R4JournalismBook/HowDoI.html How Do I? …(do that in R)] by Sharon Machlis
* [https://www.t4rstats.com/ Twitter for R programmers]
 
== Bug Tracking System ==
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
 
Use '''sessionInfo()'''.


## Method 2: mvtnorm::rmvnorm()
== License ==
library(mvtnorm)
[http://www.win-vector.com/blog/2019/07/some-notes-on-gnu-licenses-in-r-packages/ Some Notes on GNU Licenses in R Packages]
set.seed(1234)
sigma <- matrix(c(4,2,2,3), ncol=2)
x <- rmvnorm(n=n, rep(0, p), sigma=covMat)
range(x)
# Mac: [1] -4.482566  4.459236
# Linux: [1] -4.482566  4.459236


## Method 3: mvnfast::rmvn()
[https://moderndata.plot.ly/why-dash-uses-the-mit-license/ Why Dash uses the mit license (and not a copyleft gpl license)]
set.seed(1234)
x <- mvnfast::rmvn(n, rep(0, p), covMat)
range(x)
# Mac: [1] -4.323585  4.355666
# Linux: [1] -4.323585  4.355666


library(microbenchmark)
== Interview questions ==
library(MASS)
* Does R store matrices in column-major order or row-major order?
library(mvtnorm)
** 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.
library(mvnfast)
microbenchmark(v1 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "eigen"),
              v2 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "svd"),
              v3 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "chol"),
              v4 <- rmvn(n, rep(0, p), covMat),
              v5 <- mvrnorm(n, rep(0, p), Sigma = covMat))
Unit: milliseconds
expr      min        lq
v1 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "eigen") 296.55374 300.81089
v2 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "svd") 461.81867 466.98806
v3 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "chol") 118.33759 120.01829
v4 <- rmvn(n, rep(0, p), covMat)  66.64675  69.89383
v5 <- mvrnorm(n, rep(0, p), Sigma = covMat) 291.19826 294.88038
mean    median        uq      max neval  cld
306.72485 301.99339 304.46662 335.6137  100    d
478.58536 470.44085 493.89041 571.7990  100    e
125.85427 121.26185 122.21361 151.1658  100  b 
71.67996  70.52985  70.92923 100.2622  100 a   
301.88144 296.76028 299.50839 346.7049  100  c 
</syntaxhighlight>
A little more investigation shows the eigen values differ a little bit on macOS and Linux.
<syntaxhighlight lang='rsplus'>
set.seed(1234); x <- mvrnorm(n, rep(0, p), Sigma = covMat)
debug(mvrnorm)
# eS --- macOS
# eS2 -- Linux
Browse[2]> range(abs(eS$values - eS2$values))
# [1] 0.000000e+00 1.776357e-15
Browse[2]> var(as.vector(eS$vectors))
[1] 0.002000006
Browse[2]> var(as.vector(eS2$vectors))
[1] 0.001999987
Browse[2]> all.equal(eS$values, eS2$values)
[1] TRUE
Browse[2]> which(eS$values != eS2$values)
  [1]  6  7  8  9  10  11  12  13  14  20  22  23  24  25  26  27  28  29
  ...
[451] 494 495 496 497 499 500
Browse[2]> range(abs(eS$vectors - eS2$vectors))
[1] 0.0000000 0.5636919
</syntaxhighlight>
 
== rle() running length encoding ==
* https://en.wikipedia.org/wiki/Run-length_encoding
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
* https://rosettacode.org/wiki/Run-length_encoding
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package
 
== citation() ==
<syntaxhighlight lang='rsplus'>
citation()
citation("MASS")
toBibtex(citation())
</syntaxhighlight>
 
== Monitor memory usage ==
* Windows: memory.size(max=TRUE)
* Linux
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
 
References:
* [https://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
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
* [http://www.amazon.com/Cookbook-OReilly-Cookbooks-Paul-Teetor/dp/0596809158/ref=pd_sim_b_3?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB R Cookbook] by Paul Teetor
* [http://www.amazon.com/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
* [https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
: <syntaxhighlight lang='rsplus'>
# R 3.4.1
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
setwd("/tmp/efficientR/")
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# generated pdf file is located _book/_main.pdf


bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
* Explain the difference between == and === in R. Provide an example to illustrate their use.
# generated epub file is located _book/_main.epub.
** The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.  
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
# but it is OK to run in an R terminal
</syntaxhighlight>
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro


== Webinar ==
* What is the purpose of the apply() function in R? How does it differ from the for loop?
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
** The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.


== useR! ==
* Describe the concept of factors in R. How are they used in data manipulation and analysis?
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
** Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]
* [http://www.user2019.fr/ useR! 2019], [https://github.com/sowla/useR2019-materials tutorial]


== R consortium ==
* What is the significance
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
of the attach() and detach() functions in R? When should they be used?
** A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.


== Blogs, Tips, Socials, Communities ==
* Explain the concept of vectorization in R. How does it impact the performance of R code?
* Google: revolutionanalytics In case you missed it
** Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]


== Bug Tracking System ==
* Describe the difference between data.frame and matrix in R. When would you use one over the other?
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
** A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
** A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
** You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.


Use '''sessionInfo()'''.
* What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
** The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
** It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
** To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.

Revision as of 13:56, 22 April 2024

Install and upgrade R

Here

New release

Online Editor

We can run R on web browsers without installing it on local machines (similar to [/ideone.com Ideone.com] for C++. It does not require an account either (cf RStudio).

rdrr.io

It can produce graphics too. The package I am testing (cobs) is available too.

rstudio.cloud

RDocumentation

The interactive engine is based on DataCamp Light

For example, tbl_df function from dplyr package.

The website DataCamp allows to run library() on the Script window. After that, we can use the packages on R Console.

Here is a list of (common) R packages that users can use on the web.

The packages on RDocumentation may be outdated. For example, the current stringr on CRAN is v1.2.0 (2/18/2017) but RDocumentation has v1.1.0 (8/19/2016).

Web Applications

R web applications

Creating local repository for CRAN and Bioconductor

R repository

Parallel Computing

See R parallel.

Cloud Computing

Install R on Amazon EC2

http://randyzwitch.com/r-amazon-ec2/

Bioconductor on Amazon EC2

http://www.bioconductor.org/help/bioconductor-cloud-ami/

Big Data Analysis

bigmemory, biganalytics, bigtabulate

ff, ffbase

biglm

data.table

See data.table.

disk.frame

Split-apply-combine for Maximum Likelihood Estimation of a linear model

Apache arrow

Reproducible Research

Reproducible Environments

https://rviews.rstudio.com/2019/04/22/reproducible-environments/

checkpoint package

Some lessons in R coding

  1. don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use Rcpp package and R's random number generator instead.
  2. don't use list.files() directly. The result is platform dependent even different Linux OS. An extra sorting helps!

Useful R packages

Rcpp

http://cran.r-project.org/web/packages/Rcpp/index.html. See more here.

RInside : embed R in C++ code

Ubuntu

With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on Youtube. I have tested this qtdensity example successfully using Qt 4.8.5.

  1. Follow the instruction cairoDevice to install required libraries for cairoDevice package and then cairoDevice itself.
  2. Install Qt. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
  3. Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
  4. Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the interactive program qtdensity appears on your desktop.

File:qtdensity.png

With RInside + Wt web toolkit installed, we can also create a web application. To demonstrate the example in examples/wt directory, we can do

cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
make
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080

Then we can go to the browser's address bar and type http://localhost:8080 to see how it works (a screenshot is in here).

Windows 7

To make RInside works on Windows OS, try the following

  1. Make sure R is installed under C:\ instead of C:\Program Files if we don't want to get an error like g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory.
  2. Install RTools
  3. Instal RInside package from source (the binary version will give an error )
  4. Create a DOS batch file containing necessary paths in PATH environment variable
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
cmd

In the Windows command prompt, run

cd C:\R\R-3.0.1\library\RInside\examples\standard
make -f Makefile.win

Now we can test by running any of executable files that make generates. For example, rinside_sample0.

rinside_sample0

As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See some discussions in

So the Qt and Wt web tool applications on Windows may or may not be possible.

GUI

Qt and R

tkrplot

On Ubuntu, we need to install tk packages, such as by

sudo apt-get install tk-dev

reticulate - Interface to 'Python'

Python -> reticulate

Hadoop (eg ~100 terabytes)

See also HighPerformanceComputing

RHadoop

Snowdoop: an alternative to MapReduce algorithm

XML

On Ubuntu, we need to install libxml2-dev before we can install XML package.

sudo apt-get update
sudo apt-get install libxml2-dev

On CentOS,

yum -y install libxml2 libxml2-devel

XML

library(XML)

# Read and parse HTML file
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)

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

# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)

# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')

This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.

> library(RCurl) # getURL()
> library(XML)   # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # 2.5.3a
> plain.text
 [1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"     
 [6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"

> # try picard
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
 [1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
[10] "2.6.0"

This method can be used to monitor new tags/releases from some projects like Cura, BWA, Picard, STAR. But for some projects like sratools the class attribute in the span element ("css-truncate-target") can be different (such as "tag-name").

xmlview

RCurl

On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)

# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev

Scrape google scholar results

https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R

No google ID is required

Seems not work

 Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  : 
  arguments imply differing number of rows: 2, 0 

devtools

devtools package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the remotes package which was suggested by the BiocManager package.

# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev

# Ubuntu 16.04, 18.04
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev

# Ubuntu 20.04
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev

Lazy-load database XXX is corrupt. internal error -3. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.

NB. According to the output of apt-cache show r-cran-devtools, the binary package is very old though apt-cache show r-base and supported packages like survival shows the latest version.

httr

httr imports curl, jsonlite, mime, openssl and R6 packages.

When I tried to install httr package, I got an error and some message:

Configuration failed because openssl was not found. Try installing:
 * deb: libssl-dev (Debian, Ubuntu, etc)
 * rpm: openssl-devel (Fedora, CentOS, RHEL)
 * csw: libssl_dev (Solaris)
 * brew: openssl (Mac OSX)
If openssl is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’

It turns out after I run sudo apt-get install libssl-dev in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!

Real example: see this post. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).

Since httr package was used in many other packages, take a look at how others use it. For example, aRxiv package.

A package to download free Springer books during Covid-19 quarantine, An update to "An adventure in downloading books" (rvest package)

curl

curl is independent of RCurl package.

library(curl)
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)

rOpenSci packages

rOpenSci contains packages that allow access to data repositories through the R statistical programming environment

remotes

Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).

Example:

# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')

DirichletMultinomial

On Ubuntu, we do

sudo apt-get install libgsl0-dev

Create GUI

gWidgets

GenOrd: Generate ordinal and discrete variables with given correlation matrix and marginal distributions

here

json

R web -> json

Map

leaflet

choroplethr

ggplot2

How to make maps with Census data in R

googleVis

See an example from RJSONIO above.

googleAuthR

Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.

gtrendsR - Google Trends

quantmod

Maintaining a database of price files in R. It consists of 3 steps.

  1. Initial data downloading
  2. Update existing data
  3. Create a batch file

caret

Tool for connecting Excel with R

write.table

Output a named vector

vec <- c(a = 1, b = 2, c = 3)
write.csv(vec, file = "my_file.csv", quote = F)
x = read.csv("my_file.csv", row.names = 1)
vec2 <- x[, 1]
names(vec2) <- rownames(x)
all.equal(vec, vec2)

# one liner: row names of a 'matrix' become the names of a vector
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
all.equal(vec, vec3)

Avoid leading empty column to header

write.table writes unwanted leading empty column to header when has rownames

write.table(a, 'a.txt', col.names=NA)

Add blank field AND column names in write.table

  • write.table(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
    • Suppose x is (n x 2)
    • write.table(x, sep="\t") will generate a file with 2 element on the 1st row
    • read.table(file) will return an object with a size (n x 2)
    • read.delim(file) and read.delim2(file) will also be correct
  • Note that write.csv() does not have this issue that write.table() has
    • Suppose x is (n x 2)
    • Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
    • If we use read.csv(file), the object is (n x 3). So we need to use read.csv(file, row.names = 1)
  • adding blank field AND column names in write.table(); write.table writes unwanted leading empty column to header when has rownames
write.table(a, 'a.txt', col.names=NA)
  • readr::write_tsv() does not include row names in the output file

read.delim(, row.names=1) and write.table(, row.names=TRUE)

How to Use read.delim Function in R

Case 1: no row.names

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...

Case 2: with row.names. Note: if we open the text file in Excel, we'll see the 1st row is missing one header at the end. It is actually missing the column name for the 1st column.

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
my_df <- read.delim('my_data.txt')  # it will automatically assign the rownames

Read/Write Excel files package

  • http://www.milanor.net/blog/?p=779
  • flipAPI. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
  • xlsx: depends on Java
  • openxlsx: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
  • readxl: it does not depend on anything although it can only read but not write Excel files.
    • It is part of tidyverse package. The readxl website provides several articles for more examples.
    • readxl webinar.
    • One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a tbl_df, a tbl and a data.frame.
    • For writing to Excel formats, use writexl or openxlsx package.
library(readxl)
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE, 
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf, 
    guess_max = min(1000, n_max), progress = readxl_progress(), 
    .name_repair = "unique")
# Example
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
  • writexl: zero dependency xlsx writer for R
library(writexl)
mylst <- list(sheet1name = df1, sheet2name = df2)
write_xlsx(mylst, "output.xlsx")

For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).

> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                   HK1A1   21652 Hs.445981
2                   HK1A2   22012 Hs.119177
3                   HK1A4   22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1     1495
2                              ADP-ribosylation factor 3   ARF3      377
3                          Uroporphyrinogen III synthase   UROS     7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1   5.000000        5q31.2               <NA>      1
2  12.000000         12q13               <NA>      1
3       <NA> 10q25.2-q26.3               <NA>      1

The hidden worksheets become visible (Not sure what are those first rows mean in the output).

> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00 
[1] "Experiment descriptors" "Filtered log ratio"     "Gene identifiers"      
[4] "Gene annotations"       "CollateInfo"            "GeneSubsets"           
[7] "GeneSubsetsTemp"       

The Chinese character works too.

> read_excel("~/Downloads/testChinese.xlsx", 1)
   中文 B C
1     a b c
2     1 2 3

To read all worksheets we need a convenient function

read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc1) is a tibble.

readr

Compared to base equivalents like read.csv(), readr is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.

1.0.0 released. readr 2.0.0 adds built-in support for reading multiple files at once, fast multi-threaded lazy reading and automatic guessing of delimiters among other changes.

Consider a text file where the table (6100 x 22) has duplicated row names and the (1,1) element is empty. The column names are all unique.

  • read.delim() will treat the first column as rownames but it does not allow duplicated row names. Even we use row.names=NULL, it still does not read correctly. It does give warnings (EOF within quoted string & number of items read is not a multiple of the number of columns). The dim is 5177 x 22.
  • readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
  • data.table::fread(Filename, sep = "\t") will detect the number of column names is less than the number of columns. Added 1 extra default column name for the first column which is guessed to be row names or an index. The dim is 6100 x 22. (Winner!)

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

Note that data.table::fread() can read a selection of the columns.

Speed comparison

The Fastest Way To Read And Write Files In R. data.table >> readr >> base.

ggplot2

See ggplot2

Data Manipulation & Tidyverse

See Tidyverse.

Data Science

See Data science page

microbenchmark & rbenchmark

Plot, image

jpeg

If we want to create the image on this wiki left hand side panel, we can use the jpeg package to read an existing plot and then edit and save it.

We can also use the jpeg package to import and manipulate a jpg image. See Fun with Heatmaps and Plotly.

png and resolution

It seems people use res=300 as a definition of high resolution.

  • Bottom line: fix res=300 and adjust height/width as needed. The default is res=72, height=width=480. If we increase res=300, the text font size will be increased, lines become thicker and the plot looks like a zoom-in.
  • Saving high resolution plot in png.
    png("heatmap.png", width = 8, height = 6, units='in', res = 300) 
    # we can adjust width/height as we like
    # the pixel values will be width=8*300 and height=6*300 which is equivalent to 
    # 8*300 * 6*300/10^6 = 4.32 Megapixels (1M pixels = 10^6 pixels) in camera's term
    # However, if we use png(, width=8*300, height=6*300, units='px'), it will produce
    # a plot with very large figure body and tiny text font size.
    
    # It seems the following command gives the same result as above
    png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
    
  • Chapter 14.5 Outputting to Bitmap (PNG/TIFF) Files by R Graphics Cookbook
    • Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.
  • 10 tips for making your R graphics look their best David Smith
    • In Word you can resize the graphic to an appropriate size, but the high resolution gives you the flexibility to choose a size while not compromising on the quality. I'd recommend at least 1200 pixels on the longest side for standard printers.
  • ?png. The png function has default settings ppi=72, height=480, width=480, units="px".
    • By default no resolution is recorded in the file, except for BMP.
    • BMP vs PNG format. If you need a smaller file size and don’t mind a lossless compression, PNG might be a better choice. If you need to retain as much detail as possible and don’t mind a larger file size, BMP could be the way to go.
      • Compression: BMP files are raw and uncompressed, meaning they’re large files that retain as much detail as possible. On the other hand, PNG files are compressed but still lossless. This means you can reduce or expand PNGs without losing any information.
      • File size: BMPs are larger than PNGs. This is because PNG files automatically compress, and can be compressed again to make the file even smaller.
      • Common uses: BMP contains a maximum amount of details while PNGs are good for small illustrations, sketches, drawings, logos and icons.
      • Quality: No difference
      • Transparency: PNG supports transparency while BMP doesn't
  • Some comparison about the ratio
    • 11/8.5=1.29 (A4 paper)
    • 8/6=1.33 (plot output)
    • 1440/900=1.6 (my display)
  • Setting resolution and aspect ratios in R
  • The difference of res parameter for a simple plot. How to change the resolution of a plot in base R?
  • High Resolution Figures in R.
  • High resolution graphics with R
  • R plot: size and resolution
  • How can I increase the resolution of my plot in R?, devEMF package
  • See Images -> Anti-alias.
  • How to check DPI on PNG
    • The width of a PNG file in terms of inches cannot be determined directly from the file itself, as the file contains pixel dimensions, not physical dimensions. However, you can calculate the width in inches if you know the resolution (DPI, dots per inch) of the image. Remember that converting pixel measurements to physical measurements like inches involves a specific resolution (DPI), and different devices may display the same image at different sizes due to having different resolutions.
  • Cairo case.

PowerPoint

  • For PP presentation, I found it is useful to use svg() to generate a small size figure. Then when we enlarge the plot, the text font size can be enlarged too. According to svg, by default, width = 7, height = 7, pointsize = 12, family = sans.
  • Try the following code. The font size is the same for both plots/files. However, the first plot can be enlarged without losing its quality.
    svg("svg4.svg", width=4, height=4)
    plot(1:10, main="width=4, height=4")
    dev.off()
    
    svg("svg7.svg", width=7, height=7) # default
    plot(1:10, main="width=7, height=7")
    dev.off()
    

magick

https://cran.r-project.org/web/packages/magick/

See an example here I created.

Cairo

See White strips problem in png() or tiff().

geDevices

cairoDevice

PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).

For ubuntu OS, we need to install 2 libraries and 1 R package RGtk2.

sudo apt-get install libgtk2.0-dev libcairo2-dev

On Windows OS, we may got the error: unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' . We need to follow the instruction in here.

dpi requirement for publication

For import into PDF-incapable programs (MS Office)

sketcher: photo to sketch effects

https://htsuda.net/sketcher/

httpgd

igraph

creating directed networks with igraph

Identifying dependencies of R functions and scripts

https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts

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

foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)

foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)

iterators

Iterator is useful over for-loop if the data is already a collection. It can be used to iterate over a vector, data frame, matrix, file

Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.

Colors

  • scales package. This is used in ggplot2 package.
  • colorspace: A Toolbox for Manipulating and Assessing Colors and Palettes. Popular! Many reverse imports/suggests; e.g. ComplexHeatmap. See my ggplot2 page.
    hcl_palettes(plot = TRUE) # a quick overview
    hcl_palettes(palette = "Dark 2", n=5, plot = T)
    q4 <- qualitative_hcl(4, palette = "Dark 3")
    
  • convert hex value to color names
    library(plotrix)
    sapply(rainbow(4), color.id) # color.id is a function
              # it is used to identify closest match to a color
    sapply(palette(), color.id)
    sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
    

Below is an example using the option scale_fill_brewer(palette = "Paired"). See the source code at gist. Note that only set1 and set3 palettes in qualitative scheme can support up to 12 classes.

According to the information from the colorbrew website, qualitative schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.

File:GgplotPalette.svg

colortools

Tools that allow users generate color schemes and palettes

colourpicker

A Colour Picker Tool for Shiny and for Selecting Colours in Plots

eyedroppeR

Select colours from an image in R with {eyedroppeR}

rex

Friendly Regular Expressions

formatR

The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.

See also this discussion on stackoverflow talks about R code reformatting.

library(formatR)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
tidy_source("clipboard") 
# default width is getOption("width") which is 127 in my case.

Some issues

  • Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
cat("abcd",
    # This is my comment
    "defg")

will result in

> tidy_source("clipboard")
Error in base::parse(text = code, srcfile = NULL) : 
  3:1: unexpected string constant
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
3: "defg"
   ^
  • Comments appearing at the end of a line within a long complete statement won't break tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
cat("abcd"
    ,"defg"   # This is my comment
  ,"ghij")

will become

cat("abcd", "defg"  # This is my comment
, "ghij") 

Still bad!!

  • Comments appearing at the end of a line within a long complete statement breaks tidy_source() function. For example,
cat("</p>",
	"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
	ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
	                     "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"), 
	file=ExternalFileName, sep="\n", append=T)

will result in

> tidy_source("clipboard", width.cutoff=70)
Error in base::parse(text = code, srcfile = NULL) : 
  3:129: unexpected SPECIAL
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
  • width.cutoff parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on", 
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName, 
    sep = "\n", append = T)
  • It merges lines though I don't always want to do that. For example
cat("abcd"
    ,"defg"  
  ,"ghij")

will become

cat("abcd", "defg", "ghij") 

styler

https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.

Download papers

biorxivr

Search and Download Papers from the bioRxiv Preprint Server (biology)

aRxiv

Interface to the arXiv API

pdftools

aside: set it aside

An RStudio addin to run long R commands aside your current session.

Teaching

  • smovie: Some Movies to Illustrate Concepts in Statistics

Organize R research project

How to save (and load) datasets in R (.RData vs .Rds file)

How to save (and load) datasets in R: An overview

Naming convention

Efficient Data Management in R

Efficient Data Management in R. .Rprofile, renv package and dplyr package.

Text to speech

Text-to-Speech with the googleLanguageR package

Speech to text

https://github.com/ggerganov/whisper.cpp and an R package audio.whisper

Weather data

logR

https://github.com/jangorecki/logR

Progress bar

https://github.com/r-lib/progress#readme

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

cron

beepr: Play A Short Sound

https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".

utils package

https://www.rdocumentation.org/packages/utils/versions/3.6.2

tools package

Different ways of using R

Extending R by John M. Chambers (2016)

10 things R can do that might surprise you

https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/

R call C/C++

Mainly talks about .C() and .Call().

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

.Call

Be sure to add the PACKAGE parameter to avoid an error like

cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group, 
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) : 
  "standardize" not resolved from current namespace (grpreg)

NAMESPACE file & useDynLib

(From Writing R Extensions manual) Loading is most often done automatically based on the useDynLib() declaration in the NAMESPACE file, but may be done explicitly via a call to library.dynam(). This has the form

library.dynam("libname", package, lib.loc) 

library.dynam.unload()

gcc

Coping with varying `gcc` versions and capabilities in R packages

Primitive functions

Primitive Functions List

SEXP

Some examples from packages

  • sva package has one C code function

R call Fortran

Embedding R

An very simple example (do not return from shell) from Writing R Extensions manual

The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.

This example can be run by

R_HOME/bin/R CMD R_HOME/bin/exec/R

Note:

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

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

An example from Bioconductor workshop

Example: Create embed.c file. Then build the executable. Note that I don't need to create R_HOME variable.

cd 
tar xzvf 
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest

nano embed.c
# Using a single line will give an error and cannot not show the real problem.
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
# A better way is to run compile and link separately
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed

Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying R_HOME variable and including the directories used in linking R in LD_LIBRARY_PATH. This is based on the inform provided by Writing R Extensions.

export R_HOME=/home/brb/Downloads/R-3.0.2
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
./embed # No need to include R CMD in front.

Question: Create a data frame in C? Answer: Use data.frame() via an eval() call from C. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as here.

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

Create a Simple Socket Server in R

This example is coming from this paper.

Create an R function

simpleServer <- function(port=6543)
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}

Then run simpleServer(). Open another terminal and try to communicate with the server

$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.

Welcome to R!
R> summary(iris[, 3:5])
  Petal.Length    Petal.Width          Species  
 Min.   :1.000   Min.   :0.100   setosa    :50  
 1st Qu.:1.600   1st Qu.:0.300   versicolor:50  
 Median :4.350   Median :1.300   virginica :50  
 Mean   :3.758   Mean   :1.199                  
 3rd Qu.:5.100   3rd Qu.:1.800                  
 Max.   :6.900   Max.   :2.500                  

R> quit
Connection closed by foreign host.

Rserve

Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See Example from Bioconductor workshop.

See my Rserve page.

outsider

(Commercial) StatconnDcom

R.NET

rJava

Terminal

# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk

and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.

  • Create the file /etc/ld.so.conf.d/java.conf with the following entries:
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
  • And then run sudo ldconfig

Now go back to R

install.packages("rJava")

Done!

If above does not work, a simple way is by (under Ubuntu) running

sudo apt-get install r-cran-rjava

which will create new package 'default-jre' (under /usr/lib/jvm) and 'default-jre-headless'.

RCaller

RApache

Rscript, arguments and commandArgs()

Passing arguments to an R script from command lines Syntax:

$ Rscript --help
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]

Example:

args = commandArgs(trailingOnly=TRUE)
# test if there is at least one argument: if not, return an error
if (length(args)==0) {
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
} else if (length(args)==1) {
  # default output file
  args[2] = "out.txt"
}
cat("args[1] = ", args[1], "\n")
cat("args[2] = ", args[2], "\n")
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt 
# args[2] =  out.txt

Rscript, #! Shebang and optparse package

littler

Provides hash-bang (#!) capability for R

FAQs:

root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"

-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
                                              # Example: R -q -e "date()"

-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
                                              # It won't show the startup message when it is used in the command line.
                                              # Example: Rscript -e "date()"

We can install littler using two ways.

  • install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r). You need to create a soft link in order to access it globally.
  • sudo apt install littler. This will install 'r' globally; however, the installed version may be old.

After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The web version of examples does not have the TOC.

r was not meant to run interactively like R. See man r.

RInside: Embed R in C++

See RInside

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

The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.

To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.

To run any executable program, we need to specify LD_LIBRARY_PATH variable, something like

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib 

The real build process looks like (check <Makefile> for completeness)

g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include   \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0

Hello World example of embedding R in C++.

#include <RInside.h>                    // for the embedded R via RInside

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

    RInside R(argc, argv);              // create an embedded R instance 

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

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

    exit(0);
}

The above can be compared to the Hello world example in Qt.

#include <QApplication.h>
#include <QPushButton.h>

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

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

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

    return app.exec();
}

RFortran

RFortran is an open source project with the following aim:

To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.

It works only on Windows platform with Microsoft Visual Studio installed:(

Call R from other languages

C

Using R from C/C++

Error: “not resolved from current namespace” error, when calling C routines from R

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

JRI

http://www.rforge.net/JRI/

ryp2

http://rpy.sourceforge.net/rpy2.html

Create a standalone Rmath library

R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of R-admin manual.

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

Create a static library <libRmath.a> and a dynamic library <Rmath.dll>

Suppose we have downloaded R source code and build R from its source. See Build_R_from_its_source. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.

cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win

Use Rmath library in our code

set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.

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

# Force to link against the static library <libRmath.a>
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe

# Force to link against dynamic library <Rmath.dll>
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe

Test the executable program. Note that the executable program RmathEx1.exe can be transferred to and run in another computer without R installed. Isn't it cool!

c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689

Below is the cpp program <RmathEx1.cpp>.

//RmathEx1.cpp
#define MATHLIB_STANDALONE 
#include <iostream>
#include "Rmath.h"

using std::cout; using std::cin; using std::endl;

int main()
{
  double x1, x2;
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;

  cout << "Prob(Z <= " << x1 << ") = " << 
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " << 
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}

Calling R.dll directly

See Chapter 8.2.2 of R Extensions. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.

Create HTML report

ReportingTools (Jason Hackney) from Bioconductor. See Genome->ReportingTools.

htmlTable package

The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.

formattable

htmltab package

This package is NOT used to CREATE html report but EXTRACT html table.

ztable package

Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.

Create academic report

reports package in CRAN and in github repository. The youtube video gives an overview of the package.

Create pdf and epub files

# Idea:
#        knitr        pdflatex
#   rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
  • A very simple example <002-minimal.Rnw> from yihui.name works fine on linux.
git clone https://github.com/yihui/knitr-examples.git
  • <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run sudo apt-get install texlive-fonts-recommended to install missing fonts. It works!

To see a real example, check out DESeq2 package (inst/doc subdirectory). In addition to DESeq2, I also need to install DESeq, BiocStyle, airway, vsn, gplots, and pasilla packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.

Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.

# Idea:
#        knitr        pandoc
#   rmd -------> md ----------> pdf

git clone https://github.com/yihui/knitr-examples.git
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!

To create an epub file (not success yet on Windows OS, missing figures on Linux OS)

# Idea:
#        knitr        pandoc
#   rnw -------> tex ----------> markdown or epub

library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")

Convert tex to epub

kable() for tables

Create Tables In LaTeX, HTML, Markdown And ReStructuredText

Create Word report

Using the power of Word

How to go from R to nice tables in Microsoft Word

knitr + pandoc

It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.

# Idea:
#        knitr       pandoc
#   rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files

and then

FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))

Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.

Another way is

library(pander)
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")

Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:

  • A pdf file: pandoc -s report.md -t latex -o report.pdf
  • A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
  • Openoffice: pandoc report.md -o report.odt
  • Word docx: pandoc report.md -o report.docx

We can also create the epub file for reading on Kobo ereader. For example, download this file and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!

knit("example.Rmd")
pandoc("example.md", format="epub")

PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)

> pandoc("Rmd_to_Epub.md", format="epub")
executing pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
Error in (function (input, format, ext, cfg)  : conversion failed
In addition: Warning message:
running command 'pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1

pander

Try pandoc[1] with a minimal reproducible example, you might give a try to my "pander" package [2] too:

library(pander)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
            output = tempfile(), convert = 'docx')

Where the content of the "minimal.brew" file is something you might have got used to with Sweave - although it's using "brew" syntax instead. See the examples of pander [3] for more details. Please note that pandoc should be installed first, which is pretty easy on Windows.

  1. http://johnmacfarlane.net/pandoc/
  2. http://rapporter.github.com/pander/
  3. http://rapporter.github.com/pander/#examples

R2wd

Use R2wd package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.

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

This will download and install the current version of statconnDCOM

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

The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.

Convert from pdf to word

The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert

rtf

Use rtf package for Rich Text Format (RTF) Output.

xtable

Package xtable will produce html output.

print(xtable(X), type="html")

If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.

officer

  • CRAN. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.
  • The gist includes a comprehensive example that encompasses various elements such as sections, subsections, and tables. It also incorporates a detailed paragraph, along with visual representations created using base R plots and ggplots.
  • Add a line space
    doc <- body_add_par(doc, "")
    
    # Function to add n line spaces
    body_add_par_n <- function (doc, n) {
      for(i in 1:n){
        doc <- body_add_par(doc, "")
      }
      return(doc)
    }
    body_add_par_n(3)
    
  • Figures from the documentation of officeverse.
  • See Data frame to word table?.
  • See Office page for some code.
  • How to read and create Word Documents in R where we can extracting tables from Word Documents.
    x = read_docx("myfile.docx")
    content <- docx_summary(x) # a vector
    grep("nlme", content$text, ignore.case = T, value = T)
    

Powerpoint

PDF manipulation

staplr

R Graphs Gallery

COM client or server

Client

RDCOMClient where excel.link depends on it.

Server

RDCOMServer

Use R under proxy

http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy

RStudio

  • Github
  • Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
  • Preview

rstudio.cloud

https://rstudio.cloud/

Launch RStudio

Multiple versions of R

Create .Rproj file

If you have an existing package that doesn't have an .Rproj file, you can use devtools::use_rstudio("path/to/package") to add it.

With an RStudio project file, you can

  • Restore .RData into workspace at startup
  • Save workspace to .RData on exit (or save.image("Robj.RData") & load("Robj.RData"))
  • Always save history (even if no saving .RData, savehistory(".Rhistory") & loadhistory(".Rhistory"))
  • etc

package search

https://github.com/RhoInc/CRANsearcher

Git

Visual Studio

R and Python support now built in to Visual Studio 2017

List files using regular expression

  • Extension
list.files(pattern = "\\.txt$")

where the dot (.) is a metacharacter. It is used to refer to any character.

  • Start with
list.files(pattern = "^Something")

Using Sys.glob()"' as

> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"       "/home/brb/Downloads/valgrind.txt"

Hidden tool: rsync in Rtools

c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe

sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
total size is 1198416  speedup is 3.71

c:\Rtools\bin>

Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also rsync file permissions on windows.

Install rgdal package (geospatial Data) on ubuntu

Terminal

sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411

R

install.packages("rgdal")

Install sf package

I got the following error even I have installed some libraries.

checking GDAL version >= 2.0.1... no
configure: error: sf is not compatible with GDAL versions below 2.0.1

Then I follow the instruction here

sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable

sudo apt update
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears 

sudo apt install libgdal-dev # works on ubuntu 20.04 too
                             # no need the previous lines

Database

RSQLite

Creating a new database:

library(DBI)

mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
dbDisconnect(mydb)
unlink("my-db.sqlite")

# temporary database
mydb <- dbConnect(RSQLite::SQLite(), "")
dbDisconnect(mydb)

Loading data:

mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)

dbListTables(mydb)

dbListFields(con, "mtcars")

dbReadTable(con, "mtcars")

Queries:

dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))

res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
dbFetch(res)

Batched queries:

dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}

dbClearResult(rs)

Multiple parameterised queries:

rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)

Statements:

dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
#> [1] 0
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)

sqldf

Manipulate R data frames using SQL. Depends on RSQLite. A use of gsub, reshape2 and sqldf with healthcare data

RPostgreSQL

RMySQL

MongoDB

odbc

RODBC

DBI

dbplyr

Create a new SQLite database:

surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")

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

copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db

Connect to a database:

download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")

library(dbplyr)
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")

Querying the database with the SQL syntax:

tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))

Querying the database with the dplyr syntax:

surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)

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

Simple database queries:

surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)

Laziness (instruct R to stop being lazy):

data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()

Complex database queries:

plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table

surveys # The plot_id column also features in the surveys table

# Join databases method 1
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()

NoSQL

nodbi: the NoSQL Database Connector

Github

R source

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

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

R packages (only) source (metacran)

Bioconductor packages source

Announcement, https://github.com/Bioconductor-mirror

Send local repository to Github in R by using reports package

http://www.youtube.com/watch?v=WdOI_-aZV0Y

My collection

How to download

Clone ~ Download.

  • Command line
git clone https://gist.github.com/4484270.git

This will create a subdirectory called '4484270' with all cloned files there.

  • Within R
library(devtools)
source_gist("4484270")

or First download the json file from

https://api.github.com/users/MYUSERLOGIN/gists

and then

library(RJSONIO)
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})

Jekyll

An Easy Start with Jekyll, for R-Bloggers

Connect R with Arduino

Android App

Common plots tips

Create an empty plot

plot.new()

Overlay plots

How to Overlay Plots in R-Quick Guide with Example.

#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)

Save the par() and restore it

Example 1: Don't use old.par <- par() directly. no.readonly = FALSE by default. * The `no.readonly = TRUE` argument in the par() function in R is used to get the full list of graphical parameters that can be restored.

  • When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
  • If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)

Example 2: Use it inside a function with the on.exit(0 function.

ex <- function() {
   old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
   on.exit(par(old.par))
   ## ... do lots of par() settings and plots
   ## ...
   invisible() #-- now,  par(old.par)  will be executed
}

Example 3: It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.

ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar

Grouped boxplots

Weather Time Line

The plot looks similar to a boxplot though it is not. See a screenshot on Android by Sam Ruston.

Horizontal bar plot

library(ggplot2)
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA", 
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) + 
  coord_flip() + xlab("") + ylab("Fold Change")   

File:Ggplot2bar.svg

Include bar values in a barplot

Use text().

Or use geom_text() if we are using the ggplot2 package. See an example here or this.

For stacked barplot, see this post.

Grouped barplots

library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) + 
       geom_bar(position="dodge", stat="identity")
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042

Math expression

# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
     ylab = expression(phi^{rho + a}),
     main = "Pure Expressions")

# Superscript
plot(1:10, main = expression("My Title"^2)) 
# Subscript
plot(1:10, main = expression("My Title"[2]))  

# Expressions with Spacing
# '~' is to add space and '*' is to squish characters together
plot(1:10, xlab= expression(Delta * 'C'))
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
     ylab = expression(phi^{rho + a} * z * w),
     main = "Pure Expressions with Spacing")

# Expressions with Text
plot(x,y, 
     xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")), 
     ylab = expression(paste("Here is some text of ", phi^{rho})), 
     main = "Expressions with Text")

# Substituting Expressions
plot(x,y, 
     xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)), 
     ylab = substitute(paste("e is = ", e ), list(e = ee)), 
     main = "Substituted Expressions")

Impose a line to a scatter plot

  • abline + lsfit # least squares
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
  • abline + line # robust line fitting
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
  • lines
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')

How to actually make a quality scatterplot in R: axis(), mtext()

How to actually make a quality scatterplot in R

3D scatterplot

Rotating x axis labels for barplot

https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot

barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)

Set R plots x axis to show at y=0

https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0

plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")

Different colors of axis labels in barplot

See Vary colors of axis labels in R based on another variable

Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.

tN <- table(Ni <- stats::rpois(100, lambda = 5))
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")

Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.

barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)

Use text() to draw labels on X/Y-axis including rotation

par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
     adj = 1, labels = myData$names, xpd = TRUE)

Vertically stacked plots with the same x axis

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

Include labels on the top axis/margin: axis() and mtext()

plot(1:4, rnorm(4), axes = FALSE)
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3

See also 15_Questions_All_R_Users_Have_About_Plots

This can be used to annotate each plot with the script name, date, ...

mtext(text=paste("Prepared on", format(Sys.time(), "%d %B %Y at %H:%M")), 
      adj=.99,  # text align to right 
      cex=.75, side=3, las=1, line=2)

ggplot2 uses breaks instead of at parameter. See ggplot2 → Add axis on top or right hand side, ggplot2 → scale_x_continus(name, breaks, labels) and the scale_continuous documentation.

Legend tips

Add legend to a plot in R

Increase/decrease legend font size cex & ggplot2 package case.

plot(rnorm(100))
# op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)

legend inset. Default is 0. % (from 0 to 1) to draw the legend away from x and y axis. The inset argument with negative values moves the legend outside the plot.

legend("bottomright", inset=.05, )

legend without a box

legend(, bty = "n")

Add a legend title

legend(, title = "")

Add a common legend to multiple plots. Use the layout function.

Superimpose a density plot or any curves

Use lines().

Example 1

plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))

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

Example 2

require(survival)
n = 10000
beta1 = 2; beta2 = -1
lambdaT = 1 # baseline hazard
lambdaC = 2  # hazard of censoring
set.seed(1234)
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2)) 
C <- rweibull(n, shape=1, scale=lambdaC)   
time = pmin(T,C)  
status <- 1*(T <= C) 
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1), 
     ylab="Survival probability",
     main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull

Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".

log scale

If we set y-axis to use log-scale, then what we display is the value log(Y) or log10(Y) though we still label the values using the input. For example, when we plot c(1, 10, 100) using the log scale, it is like we draw log10(c(1, 10, 100)) = c(0,1,2) on the plot but label the axis using the true values c(1, 10, 100).

File:Logscale.png

Custom scales

Using custom scales with the 'scales' package

Time series

Time series stock price plot

library(quantmod)
getSymbols("AAPL")
getSymbols("IBM") # similar to AAPL
getSymbols("CSCO") # much smaller than AAPL, IBM
getSymbols("DJI") # Dow Jones, huge 
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)", 
    col='orange', subset = '2017::2017-08')

tail(Cl(DJI))

tidyquant: Getting stock data

The 'largest stock profit or loss' puzzle: efficient computation in R

Timeline plot

Clockify

Clockify

Circular plot

Word cloud

Text mining

World map

Visualising SSH attacks with R (rworldmap and rgeolocate packages)

Diagram/flowchart/Directed acyclic diagrams (DAGs)

DiagrammeR

diagram

Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams

DAGitty (browser-based and R package)

dagR

Gmisc

Easiest flowcharts eveR?

Concept Maps

concept-maps where the diagrams are generated from https://app.diagrams.net/.

flow

flow, How To Draw Flow Diagrams In R

Venn Diagram

Venn diagram

hexbin plot

Bump chart/Metro map

https://dominikkoch.github.io/Bump-Chart/

Amazing/special plots

See Amazing plot.

Google Analytics

GAR package

http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html

Linear Programming

http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/

Linear Algebra

Amazon Alexa

R and Singularity

https://rviews.rstudio.com/2017/03/29/r-and-singularity/

Teach kids about R with Minecraft

http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html

Secure API keys

Securely store API keys in R scripts with the "secret" package

Credentials and secrets

How to manage credentials and secrets safely in R

Hide a password

keyring package

getPass

getPass

Vision and image recognition

Creating a Dataset from an Image

Creating a Dataset from an Image in R Markdown using reticulate

Turn pictures into coloring pages

https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0

Numerical optimization

CRAN Task View: Numerical Mathematics, CRAN Task View: Optimization and Mathematical Programming

Ryacas: R Interface to the 'Yacas' Computer Algebra System

Doing Maths Symbolically: R as a Computer Algebra System (CAS)

Game

Music

  • gm. Require to install MuseScore, an open source and free notation software.

SAS

sasMap Static code analysis for SAS scripts

R packages

R packages

Tricks

Getting help

Better Coder/coding, best practices

E-notation

6.022E23 (or 6.022e23) is equivalent to 6.022×10^23

Getting user's home directory

See What are HOME and working directories?

# Windows
normalizePath("~")   # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")   # "C:/Users/brb/Documents"

# Mac
normalizePath("~")   # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # "/Users/brb"

# Linux
normalizePath("~")   # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # [1] "/home/brb"

tempdir()

The path is a per-session temporary directory. On parallel use, R processes forked by functions such as mclapply and makeForkCluster in package parallel share a per-session temporary directory.

Distinguish Windows and Linux/Mac, R.Version()

identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.

get_os <- function(){
  sysinf <- Sys.info()
  if (!is.null(sysinf)){
    os <- sysinf['sysname']
    if (os == 'Darwin')
      os <- "osx"
  } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
names(R.Version())
#  [1] "platform"       "arch"           "os"             "system"        
#  [5] "status"         "major"          "minor"          "year"          
#  [9] "month"          "day"            "svn rev"        "language"      
# [13] "version.string" "nickname" 
getRversion()
# [1] ‘4.3.0’

Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only)

If we like to install R packages to a personal directory, follow this. Just add the line

R_LIBS_SITE=F:/R/library

to the file R_HOME/etc/x64/Renviron.site. In R, run Sys.getenv("R_LIBS_SITE") or Sys.getenv("R_LIBS_USER") to query the environment variable. See Environment Variables.

What is the best place to save Rconsole on Windows platform

Put/create the file <Rconsole> under C:/Users/USERNAME/Documents folder so no matter how R was upgraded/downgraded, it always find my preference.

My preferred settings:

  • Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
  • Size: 12
  • background: black
  • normaltext: white
  • usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow

and others (default options)

  • pagebg: white
  • pagetext: navy
  • highlight: DarkRed
  • dataeditbg: white
  • dataedittext: navy (View() function)
  • dataedituser: red
  • editorbg: white (edit() function)
  • editortext: black

A copy of the Rconsole is saved in github.

How R starts up

https://rstats.wtf/r-startup.html

startup - Friendly R Startup Configuration

https://github.com/henrikbengtsson/startup

Saving and loading history automatically: .Rprofile & local()

  • savehistory("filename"). It will save everything from the beginning to the command savehistory() to a text file.
  • .Rprofile will automatically be loaded when R has started from that directory
  • Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See Project-oriented workflow.
  • .Rprofile has been created/used by the packrat package to restore a packrat environment. See the packrat/init.R file and R packages → packrat.
  • Customizing Startup from R in Action, Fun with .Rprofile and customizing R startup
    • You can also place a .Rprofile file in any directory that you are going to run R from or in the user home directory.
    • At startup, R will source the Rprofile.site file. It will then look for a .Rprofile file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
    options(continue="  ") # default is "+ "
    options(prompt="R> ", continue=" ")
    options(editor="nano") # default is "vi" on Linux
    # options(htmlhelp=TRUE) 
    
    local({r <- getOption("repos")
          r["CRAN"] <- "https://cran.rstudio.com"
          options(repos=r)})
    
    .First <- function(){
     # library(tidyverse)
     cat("\nWelcome at", date(), "\n")
    }
    
    .Last <- function(){
     cat("\nGoodbye at ", date(), "\n")
    }  
    
  • https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
  • The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define R_HISTFILE system variable.
  • local() function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)

Linux or Mac

In ~/.profile or ~/.bashrc I put:

export R_HISTFILE=~/.Rhistory

In ~/.Rprofile I put:

if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory")) 
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Windows

If you launch R by clicking its icon from Windows Desktop, the R starts in C:\User\$USER\Documents directory. So we can create a new file .Rprofile in this directory.

if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Disable "Save workspace image?" prompt when exit R?

How to disable "Save workspace image?" prompt in R?

R release versions

rversions: Query the main 'R' 'SVN' repository to find the released versions & dates.

getRversion()

getRversion()
[1] ‘4.3.0’

Detect number of running R instances in Windows

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
INFO: No tasks are running which match the specified criteria.

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

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rgui.exe                      1096 Console                    1     44,712 K

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

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rserve.exe                    6108 Console                    1    381,796 K

In R, we can use

> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
[1] ""                                                                            
[2] "Image Name                     PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1     44,804 K"

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

Editor

http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs

  • Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
    • Edit the file C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el with something like
    (setq-default inferior-R-program-name
                  "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
    

GUI for Data Analysis

Update to Data Science Software Popularity 6/7/2023

BlueSky Statistics

Rcmdr

http://cran.r-project.org/web/packages/Rcmdr/index.html. After loading a dataset, click Statistics -> Fit models. Then select Linear regression, Linear model, GLM, Multinomial logit model, Ordinal regression model, Linear mixed model, and Generalized linear mixed model. However, Rcmdr does not include, e.g. random forest, SVM, glmnet, et al.

Deducer

http://cran.r-project.org/web/packages/Deducer/index.html

jamovi

Scope

See

source()

## foo.R ##
cat(ArrayTools, "\n")
## End of foo.R

# 1. Error
predict <- function() {
  ArrayTools <- "C:/Program Files" # or through load() function 
  source("foo.R")                  # or through a function call; foo()
}
predict()   # Object ArrayTools not found

# 2. OK. Make the variable global
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict()  
ArrayTools

# 3. OK. Create a global variable
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()

Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.

Example 1.

> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a      N5
2    b     999
3    c     999
4    d     999
5    e     999
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999

Example 2. How can we set global variables inside a function? The answer is to use the "<<-" operator or assign(, , envir = .GlobalEnv) function.

Other resource: Advanced R by Hadley Wickham.

Example 3. Writing functions in R, keeping scoping in mind

New environment

Run the same function on a bunch of R objects

mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(mye[[n]])

Just look at the contents of rda file without saving to anywhere (?load)

local({
   load("myfile.rda")
   ls()
})

Or use attach() which is a wrapper of load(). It creates an environment and slots it into the list right after the global environment, then populates it with the objects we're attaching.

attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
ls(pos = 2)
##  also typically need to cleanup the search path:
detach("file:all.rda")

If we want to read data from internet, load() works but not attach().

con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840

source() case.

myEnv <- new.env()    
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))

str( , max) function

Use max.level parameter to avoid a long display of the structure of a complex R object. Use give.head = FALSE to hide the attributes. See ?str

If we use str() on a function like str(lm), it is equivalent to args(lm)

For a complicated list object, it is useful to use the max.level argument; e.g. str(, max.level = 1)

For a large data frame, we can use the tibble() function; e.g. mydf %>% tibble()

tidy() function

broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See here.

View all objects present in a package, ls()

https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get

> devtools::load_all("mypackage")
> search()
 [1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
 [4] "package:stats"     "package:graphics"  "package:grDevices"
 [7] "package:utils"     "package:datasets"  "package:methods"
[10] "Autoloads"         "package:base"

> ls("package:mypackage")
[1] "_mypackage_rcpp_hello_world" "evalCpp"                     "library.dynam.unload"       
[4] "rcpp_hello_world"            "system.file"

Note that the first argument of ls() (or detach()) is used to specify the environment. It can be

  • an integer (the position in the ‘search’ list);
  • the character string name of an element in the search list;
  • an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).

Speedup R code

Profiler

&& vs &

See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.

  • The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
  • The longer form evaluates left to right examining only the first element of each vector. The return is one value.
  • The longer form evaluates left to right examining only the first element of each vector. Evaluation proceeds only until the result is determined.
  • The idea of the longer form && in R seems to be the same as the && operator in linux shell; see here.
  • Single or double?: AND operator and OR operator in R. The confusion might come from the inconsistency when choosing these operators in different languages. For example, in C, & performs bitwise AND, while && does Boolean logical AND.
  • Think of && as a stricter &
c(T,F,T) & c(T,T,T)
# [1]  TRUE FALSE  TRUE
c(T,F,T) && c(T,T,T)
# [1] TRUE
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated

It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.

> foo <- function(arg, L) {
   # Suppose 'L' is meaningful only if 'arg' is provided
   # 
   # Evaluate 'L' only if 'arg' is provided
   #
   if (!missing(arg) && L) {
     print("L is true")
   } else {
     print("Either arg is missing or L is FALSE")
   }
 }
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"

Other examples: && is more flexible than &.

nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))

if (!is.null(exprTest) && any(is.na(exprTest))) { ... }

for-loop, control flow

Vectorization

sapply vs vectorization

Speed test: sapply vs vectorization

lapply vs for loop

split() and sapply()

split() can be used to split a vector, columns or rows. See How to split a data frame?

  • Split divides the data in the vector or data frame x into the groups defined by f. The syntax is
split(x, f, drop = FALSE, …)
  • Split a vector into chunks. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the split() + lapply() + do.call() or split() + sapply() operations.
d <- 1:10
chunksize <- 4
ceiling(1:10/4)
# [1] 1 1 1 1 2 2 2 2 3 3
split(d, ceiling(seq_along(d)/chunksize))
# $`1`
# [1] 1 2 3 4
#
# $`2`
# [1] 5 6 7 8
#
# $`3`
# [1]  9 10
do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) ) 
#  1  2  3 
# 10 26 19

# bigmemory vignette
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                     function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
  • Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
split(mtcars,mtcars$cyl)

split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))   # split.data.frame() works for matrices
  • Split columns of a data frame/matrix.
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
  • split() + sapply() to merge columns. See below Mean of duplicated columns for more detail.
  • split() + sapply() to split a vector. See nsFilter() function which can remove duplicated probesets/rows using unique Entrez Gene IDs (genefilter package). The source code of nsFilter() and findLargest().
tSsp = split.default(testStat, lls) 
# testStat is a vector of numerics including probeset IDs as names
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.

sapply(tSsp, function(x) names(which.max(x))) 
# return a vector of probset IDs of length of unique entrez IDs

strsplit and sapply

> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
> strsplit(namedf, " ")
1
[1] "John" "ABC" 

2
[1] "Mary" "CDE" 

3
[1] "Kat" "FGH"

> sapply(strsplit(namedf, " "), "[", 1)
[1] "John" "Mary" "Kat" 
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"

Mean of duplicated columns: rowMeans; compute Means by each row

  • Reduce columns of a matrix by a function in R. To use rowMedians() instead of rowMeans(), we need to install matrixStats from CRAN.
    set.seed(1)
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
    colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
    res <- sapply(split(1:ncol(x), colnames(x)), 
                  function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
    res  # notice the sorting of columns
           a  b  c
     [1,] 46  1 31
     [2,] 47 12 32
     [3,] 48 13 33
     [4,] 49 14 34
     [5,] 50 15 35
     [6,] 51 16 36
     [7,] 52 17 37
     [8,] 53 18 38
     [9,] 54 19 39
    [10,] 55 20 40
    
    # vapply() is safter than sapply(). 
    # The 3rd arg in vapply() is a template of the return value.
    res2 <- vapply(split(1:ncol(x), colnames(x)), 
                   function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
                   rep(0, nrow(x)))
  • colSums, rowSums, colMeans, rowMeans (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster.
    rowMeans(x, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
    apply(x, 1, mean, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
  • matrixStats: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
  • From for() loops to the split-apply-combine paradigm for column-wise tasks: the transition for a dinosaur

Mean of duplicated rows: colMeans and rowsum

  • colMeans(x, na.rm = FALSE, dims = 1), take mean per columns & sum over rows. It returns a vector. Other similar idea functions include colSums, rowSums, rowMeans.
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
    rownames(x) <- c(rep("b", 2), rep("c", 3), rep("d", 4), "a") # move 'a' to the last
    res <- sapply(split(1:nrow(x), rownames(x)), 
                  function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
    res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
    res  # still a matrix, rows are ordered
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5 12.0 22.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    table(rownames(x))
    # a b c d
    # 1 2 3 4
    
    aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
    #   Group.1   V1   V2   V3   V4   V5   V6
    # 1       a 10.0 20.0 30.0 40.0 50.0 60.0
    # 2       b  1.5 12.0 22.0 31.5 41.5 51.5
    # 3       c  4.0 14.0 24.0 34.0 44.0 54.0
    # 4       d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays
  • rowsum(x, group, reorder = TRUE, …). Sum over rows. It returns a matrix. This is very special. It's not the same as rowSums. There is no "colsum" function. It has the speed advantage over sapply+colSums OR aggregate.
    group <- rownames(x)
    rowsum(x, group, na.rm=T)/as.vector(table(group))
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5  6.0 11.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • by() function. Calculating change from baseline in R
  • See aggregate Function in R- A powerful tool for data frames & summarize in r, Data Summarization In R
  • aggregate() function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. Don't use aggregate post.
    > attach(mtcars)
    dim(mtcars)
    [1] 32 11
    > head(mtcars)
                       mpg cyl disp  hp drat    wt  qsec vs am gear carb
    Mazda RX4         21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
    Mazda RX4 Wag     21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
    Datsun 710        22.8   4  108  93 3.85 2.320 18.61  1  1    4    1
    Hornet 4 Drive    21.4   6  258 110 3.08 3.215 19.44  1  0    3    1
    Hornet Sportabout 18.7   8  360 175 3.15 3.440 17.02  0  0    3    2
    Valiant           18.1   6  225 105 2.76 3.460 20.22  1  0    3    1
    > with(mtcars, table(cyl, vs))
       vs
    cyl  0  1
      4  1 10
      6  3  4
      8 14  0
    > aggdata <-aggregate(mtcars, by=list(cyl,vs),  FUN=mean, na.rm=TRUE)
    > print(aggdata)
      Group.1 Group.2      mpg cyl   disp       hp     drat       wt     qsec vs
    1       4       0 26.00000   4 120.30  91.0000 4.430000 2.140000 16.70000  0
    2       6       0 20.56667   6 155.00 131.6667 3.806667 2.755000 16.32667  0
    3       8       0 15.10000   8 353.10 209.2143 3.229286 3.999214 16.77214  0
    4       4       1 26.73000   4 103.62  81.8000 4.035000 2.300300 19.38100  1
    5       6       1 19.12500   6 204.55 115.2500 3.420000 3.388750 19.21500  1
             am     gear     carb
    1 1.0000000 5.000000 2.000000
    2 1.0000000 4.333333 4.666667
    3 0.1428571 3.285714 3.500000
    4 0.7000000 4.000000 1.500000
    5 0.0000000 3.500000 2.500000
    > detach(mtcars)
    
    # Another example: select rows with a minimum value from a certain column (yval in this case)
    > mydf <- read.table(header=T, text='
     id xval yval
     A 1  1
     A -2  2
     B 3  3
     B 4  4
     C 5  5
     ')
    > x = mydf$xval
    > y = mydf$yval
    > aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
      id xval yval
    1  A    1    1
    2  B    3    3
    3  C    5    5
    

Mean by Group

Mean by Group in R (2 Examples) | dplyr Package vs. Base R

aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                           # Specify function (i.e. mean)
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                         # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
               list(name = mean))               # Specify function
  • ave(x, ..., FUN),
  • aggregate(x, by, FUN),
  • by(x, INDICES, FUN): return is a list
  • tapply(): return results as a matrix or array. Useful for ragged array.

Apply family

Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.

The following list gives a hierarchical relationship among these functions.

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

Difference between apply vs sapply vs lapply vs tapply?

  • apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
  • lapply - When you want to apply a function to each element of a list in turn and get a list back.
  • sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
  • tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.

Some short examples:

Apply vs for loop

Note that, apply's performance is not always better than a for loop. See

Progress bar

What is the cost of a progress bar in R?

The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out this post. It uses winProgressBar() and setWinProgressBar() functions.

e-Rum 2020 Slides on Progressr by Henrik Bengtsson. progressr 0.8.0: RStudio's progress bar, Shiny progress updates, and absolute progress, progressr 0.10.1: Plyr Now Supports Progress Updates also in Parallel

simplify option in sapply()

library(KEGGREST)

names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2

names3 <- keggGet(c("hsa05340"))
names4 <- sapply(names3, function(x) x$GENE)
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify 
# [1] 1

lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists

  • mapply() documentation. Use mapply() to merge lists.
    mapply(rep, 1:4, 4:1)
    mapply(rep, times = 1:4, x = 4:1)
    mapply(function(x, y) seq_len(x) + y,
           c(a =  1, b = 2, c = 3),  # names from first
           c(A = 10, B = 0, C = -10))
    mapply(c, firstList, secondList, SIMPLIFY=FALSE)
    
  • Finding the Expected value of the maximum of two Bivariate Normal variables with simulation sapply + mapply.
    z <- mapply(function(u, v) { max(u, v) }, 
                u = x[, 1], v = x[, 2])
    
  • Map() and Reduce() in functional programming
  • Map(), Reduce(), and Filter() from Advanced R by Hadley
    • If you have two or more lists (or data frames) that you need to process in parallel, use Map(). One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to mapply() function and is more concise than lapply(). Advanced R has a comment that Map() is better than mapply().
      # Syntax: Map(f, ...)
      
      xs <- replicate(5, runif(10), simplify = FALSE)
      ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
      Map(weighted.mean, xs, ws)
      
      # instead of a more clumsy way
      lapply(seq_along(xs), function(i) {
        weighted.mean(xsi, wsi)
      })
      
    • Reduce() reduces a vector, x, to a single value by recursively calling a function, f, two arguments at a time. A good example of using Reduce() function is to read a list of matrix files and merge them. See How to combine multiple matrix frames into one using R?
      # Syntax: Reduce(f, x, ...)
      
      > m1 <- data.frame(id=letters[1:4], val=1:4)
      > m2 <- data.frame(id=letters[2:6], val=2:6)
      > merge(m1, m2, "id", all = T)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      > m <- list(m1, m2)
      > Reduce(function(x,y) merge(x,y, "id",all=T), m)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      

sapply & vapply

See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up this example.

rapply - recursive version of lapply

replicate

https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r

> replicate(5, rnorm(3))
           [,1]       [,2]       [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477

See parSapply() for a parallel version of replicate().

Vectorize

> rep(1:4, 4:1)
 [1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
1
[1] 1 1 1 1

2
[1] 2 2 2

3
[1] 3 3

4
[1] 4
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2)) 
[1] 1.6491761 0.9610109
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15

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

plyr and dplyr packages

Practical Data Science for Stats - a PeerJ Collection

The Split-Apply-Combine Strategy for Data Analysis (plyr package) in J. Stat Software.

A quick introduction to plyr with a summary of apply functions in R and compare them with functions in plyr package.

  1. plyr has a common syntax -- easier to remember
  2. plyr requires less code since it takes care of the input and output format
  3. plyr can easily be run in parallel -- faster

Tutorials

Examples of using dplyr:

tibble

Tibbles are data frames, but slightly tweaked to work better in the tidyverse.

Tibble objects

  • it does not have row names (cf data frame),
  • it never changes the type of the inputs (e.g. it never converts strings to factors!),
  • it never changes the names of variables

Tibbles Vignette

> data(pew, package = "efficient")
> dim(pew) 
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"     "tbl"        "data.frame"

> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
# A tibble: 162 x 3
                                                       religion Income Count
                                                          <chr>  <chr> <int>
 1                                                     Agnostic  <$10k    27
 2                                                      Atheist  <$10k    12
 ...
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
[1] NA
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count,  :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)3)
[1] 181.6975

To show all rows of a tibble object, use the print() method.

print(tbObj, n= Inf)

tbObj %>% print(n= nrow(.))

If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.

Subsetting: to extract a column from a tibble object, use [[ or $ or dplyr::pull(). Select Data Frame Columns in R.

TibbleObject$VarName
# OR
TibbleObject"VarName"
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore

dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# OR
dplyr::select(TibbleObject, 2:5) # 

Convert a data frame to a tibble See Tibble Data Format in R: Best and Modern Way to Work with Your Data

my_data <- as_tibble(iris)
class(my_data)

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

llply()

llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.

LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])

where rLLID is a list of entrez ID. For example,

get("org.Hs.egGO")[["6772"]]

returns a list of 49 GOs.

ddply()

http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html

ldply()

An R Script to Automatically download PubMed Citation Counts By Year of Publication

Performance/speed comparison

Performance comparison of converting list to data.frame with R language

Using R's set.seed() to set seeds for use in C/C++ (including Rcpp)

http://rorynolan.rbind.io/2018/09/30/rcsetseed/

get_seed()

See the same blog

get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}

Note: .Machine$integer.max = 2147483647 = 2^31 - 1.

Random seeds

By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See ?Random.

set.seed(as.numeric(Sys.time()))

set.seed(as.numeric(Sys.Date()))  # same seed for each day

.Machine and the largest integer, double

See ?.Machine.

                          Linux/Mac  32-bit Windows 64-bit Windows
double.eps              2.220446e-16   2.220446e-16   2.220446e-16
double.neg.eps          1.110223e-16   1.110223e-16   1.110223e-16
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
double.base             2.000000e+00   2.000000e+00   2.000000e+00
double.digits           5.300000e+01   5.300000e+01   5.300000e+01
double.rounding         5.000000e+00   5.000000e+00   5.000000e+00
double.guard            0.000000e+00   0.000000e+00   0.000000e+00
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
double.exponent         1.100000e+01   1.100000e+01   1.100000e+01
double.min.exp         -1.022000e+03  -1.022000e+03  -1.022000e+03
double.max.exp          1.024000e+03   1.024000e+03   1.024000e+03
integer.max             2.147484e+09   2.147484e+09   2.147484e+09
sizeof.long             8.000000e+00   4.000000e+00   4.000000e+00
sizeof.longlong         8.000000e+00   8.000000e+00   8.000000e+00
sizeof.longdouble       1.600000e+01   1.200000e+01   1.600000e+01
sizeof.pointer          8.000000e+00   4.000000e+00   8.000000e+00

NA when overflow

tmp <- 156287L
tmp*tmp
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647

How to select a seed for simulation or randomization

set.seed() allow alphanumeric seeds

https://stackoverflow.com/a/10913336

set.seed(), for loop and saving random seeds

  • Detect When the Random Number Generator Was Used
    if (interactive()) {
      invisible(addTaskCallback(local({
        last <- .GlobalEnv$.Random.seed
        
        function(...) {
          curr <- .GlobalEnv$.Random.seed
          if (!identical(curr, last)) {
            msg <- "NOTE: .Random.seed changed"
            if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
            message(msg)
            last <<- curr
          }
          TRUE
        }
      }), name = "RNG tracker"))
    }
    
  • http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
    set.seed(1001) 
    data <- vector("list", 30) 
    seeds <- vector("list", 30) 
    for(i in 1:30) { 
      seeds[[i]] <- .Random.seed 
      data[[i]] <- runif(5) 
    } 
     
    # If we save and load .Random.seed from a file using scan(), make
    # sure to convert its type from doubles to integers.
    # Otherwise, .Random.seed will complain!
    
    .Random.seed <- seeds[[23]]  # restore 
    data.23 <- runif(5) 
    data.23 
    data[[23]] 
    
  • impute.knn
  • Duncan Murdoch: This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed. See ?.Random.seed for details.
  • Uwe Ligges's comment: set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.
  • Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
  • Local randomness in R.

sample()

sample() inaccurate on very large populations, fixed in R 3.6.0

# R 3.5.3
set.seed(123)
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1 
# 400070 599930 
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

# R 3.6.0
# docker run --net=host -it --rm r-base:3.6.0
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

Getting different results with set.seed() in RStudio

Getting different results with set.seed(). It's possible that you're loading an R package that is changing the requested random number generator; RNGkind().

dplyr::sample_n()

The function has a parameter weight. For example if we have some download statistics for each day and we want to do sampling based on their download numbers, we can use this function.

Regular Expression

See here.

Read rrd file

on.exit()

Examples of using on.exit(). In all these examples, add = TRUE is used in the on.exit() call to ensure that each exit action is added to the list of actions to be performed when the function exits, rather than replacing the previous actions.

  • Database connections
    library(RSQLite)
    sqlite_get_query <- function(db, sql) {
      conn <- dbConnect(RSQLite::SQLite(), db)
      on.exit(dbDisconnect(conn), add = TRUE)
      dbGetQuery(conn, sql)
    }
    
  • File connections
    read_chars <- function(file_name) {
      conn <- file(file_name, "r")
      on.exit(close(conn), add = TRUE)
      readChar(conn, file.info(file_name)$size)
    }
    
  • Temporary files
    history_lines <- function() {
      f <- tempfile()
      on.exit(unlink(f), add = TRUE)
      savehistory(f)
      readLines(f, encoding = "UTF-8")
    }
    
  • Printing messages
    myfun = function(x) {
      on.exit(print("first"))
      on.exit(print("second"), add = TRUE)
      return(x)
    }
    

file, connection

  • cat() and scan() (read data into a vector or list from the console or file)
  • read() and write()
  • read.table() and write.table()
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)

foo <- function() {
  con <- file()
  ...
  on.exit(close(con))
  ...
}

Error in close.connection(f) : invalid connection. If we want to use close(con), we have to specify how to open the connection; such as

con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)

withr package

https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by languageserver.

Clipboard (?connections), textConnection(), pipe()

  • On Windows, we can use readClipboard() and writeClipboard().
    source("clipboard")
    read.table("clipboard")
    
  • Clipboard -> R. Reading/writing clipboard on macOS. Use textConnection() function:
    x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
    # Or on Mac
    x <- read.delim(pipe("pbpaste"))
    # safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
    

    An example is to copy data from this post. In this case we need to use read.table() instead of read.delim().

  • R -> clipboard on Mac. Note: pbcopy and pbpaste are macOS terminal commands. See pbcopy & pbpaste: Manipulating the Clipboard from the Command Line.
    • pbcopy: takes standard input and places it in the clipboard buffer
    • pbpaste: takes data from the clipboard buffer and writes it to the standard output
    clip <- pipe("pbcopy", "w")
    write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
    # write.table(data.frame(Var1, Var2), file = clip, row.names=F, quote=F, sep="\t")
    close(clip)
    
  • Clipboard -> Excel.
    • Method 1: Paste icon -> Text import wizard -> Delimit (Tab, uncheck Space) or Fixed width depending on the situation -> Finish.
    • Method 2: Ctrl+v first. Then choose Data -> Text to Columns. Fixed width -> Next -> Next -> Finish.
  • On Linux, we need to install "xclip". See R Copy from Clipboard in Ubuntu Linux. It seems to work.
    # sudo apt-get install xclip
    read.table(pipe("xclip -selection clipboard -o",open="r"))
    

clipr

clipr: Read and Write from the System Clipboard

read/manipulate binary data

  • x <- readBin(fn, raw(), file.info(fn)$size)
  • rawToChar(x[1:16])
  • See Biostrings C API

String Manipulation

format(): padding with zero

ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"

noquote()

noqute Print character strings without quotes.

stringr package

glue package

  • glue. Useful in a loop and some function like ggtitle() or ggsave().
    library(glue)
    name <- "Fred"
    glue('My name is {name}.')  # My name is Fred.
    
  • String interpolation

Raw data type

Fun with strings, Cyrillic alphabets

a1 <- "А"
a2 <- "A"
a1 == a2
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41

number of characters limit

It's a limit on a (single) input line in the REPL

HTTPs connection

HTTPS connection becomes default in R 3.2.2. See

R 3.3.2 patched The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)

setInternet2

There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.

Read the discussion reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in this post. The following demonstrated the original problem.

url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)

It seems the bug was fixed in R 3.2-branch. See 8/16/2015 patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to InternetOpenUrl() function of wininet library. This article and this post explain differences of active and passive FTP.

The following R command will show the exact svn revision for the R you are currently using.

R.Version()$"svn rev"

If setInternet2(T), then https protocol is supported in download.file().

When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.

The setInternet2() function is defined in R> src> library> utils > R > windows > sysutils.R.

R up to 3.2.2

setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))

See also

  • <src/include/Internal.h> (declare do_setInternet2()),
  • <src/main/names.c> (show do_setInternet2() in C)
  • <src/main/internet.c> (define do_setInternet2() in C).

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

R 3.3.0

setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}

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

Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan()

In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, Inf and NaN as input or output.

See ?is.finite.

How to replace Inf with NA in All or Specific Columns of the Data Frame

replace() function

File/path operations

  • list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
  • file.info()
  • dir.create()
  • file.create()
  • file.copy()
  • file.exists()
  • basename() - remove the parent path, dirname() - returns the part of the path up to but excluding the last path separator
    > file.path("~", "Downloads")
    [1] "~/Downloads"
    > dirname(file.path("~", "Downloads"))
    [1] "/home/brb"
    > basename(file.path("~", "Downloads"))
    [1] "Downloads"
    
  • path.expand("~/.Renviron") # "/home/brb/.Renviron"
  • normalizePath() # Express File Paths in Canonical Form
    > cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
    /usr/lib/R
    /tmp/RtmpzvDhAe
    
  • system.file() - Finds the full file names of files in packages etc
    > system.file("extdata", "ex1.bam", package="Rsamtools")
    [1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
    

read/download/source a file from internet

Simple text file http

retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)

Zip, RData, gz file and url() function

x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)

Here url() function is like file(), gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.

Another example is Read gzipped csv directly from a url in R

con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
                       "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))

Another example of using url() is

load(url("http:/www.example.com/example.RData"))

This does not work with load(), dget(), read.table() for files on OneDrive. In fact, I cannot use wget with shared files from OneDrive. The following trick works: How to configure a OneDrive file for use with wget.

Dropbox is easy and works for load(), wget, ...

R download .RData or Directly loading .RData from github from Github.

zip function

This will include 'hallmarkFiles' root folder in the files inside zip.

zip(zipfile = 'myFile.zip', 
    files = dir('hallmarkFiles', full.names = TRUE))

# Verify/view the files. 'list = TRUE' won't extract 
unzip('testZip.zip', list = TRUE) 

downloader package

This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.

Google drive file based on https using RCurl package

require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))

Google sheet file using googlesheets package

Reading data from google sheets into R

Github files https using RCurl package

x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt", 
            ssl.verifypeer = FALSE)
read.table(text=x)

data summary table

summarytools: create summary tables for vectors and data frames

https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.

skimr: A frictionless, pipeable approach to dealing with summary statistics

skimr for useful and tidy summary statistics

modelsummary

modelsummary: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready

broom

Tidyverse->broom

Create publication tables using tables package

See p13 for example at here

R's tables packages is the best solution. For example,

> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                  
                Sepal.Length      Sepal.Width     
 Species    n   mean         sd   mean        sd  
 setosa      50 5.01         0.35 3.43        0.38
 versicolor  50 5.94         0.52 2.77        0.31
 virginica   50 6.59         0.64 2.97        0.32
 All        150 5.84         0.83 3.06        0.44
> str(iris)
'data.frame':   150 obs. of  5 variables:
 $ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
 $ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
 $ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
 $ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
 $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...

and

# This example shows some of the less common options         
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
> z <- rnorm(100)+5
> fmt <- function(x) {
  s <- format(x, digits=2)
  even <- ((1:length(s)) %% 2) == 0
  s[even] <- sprintf("(%s)", s[even])
  s
}
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
                  Status              
 Sex    Statistic high   low    medium
 Female mean       4.88   4.96   5.17 
        sd        (1.20) (0.82) (1.35)
 Male   mean       4.45   4.31   5.05 
        sd        (1.01) (0.93) (0.75)

fgsea example

vignette & source code

(archived) ClinReport: Statistical Reporting in Clinical Trials

https://cran.r-project.org/web/packages/ClinReport/index.html

Append figures to PDF files

How to append a plot to an existing pdf file. Hint: use the recordPlot() function.

Save base graphics as pseudo-objects

Save base graphics as pseudo-objects in R. Note there are some cons with this approach.

pdf(NULL)
dev.control(displaylist="enable")
plot(df$x, df$y)
text(40, 0, "Random")
text(60, 2, "Text")
lines(stats::lowess(df$x, df$y))
p1.base <- recordPlot()
invisible(dev.off())

# Display the saved plot
grid::grid.newpage()
p1.base

Extracting tables from PDFs

Print tables

addmargins()

tableone

Some examples

Cox models

finalfit package

summary_factorlist() from the finalfit package.

table1

gtsummary

gt

Introduction to Clinical Tables with the {gt} Package

dplyr

https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.

tables::tabular()

gmodels::CrossTable()

https://www.statmethods.net/stats/frequencies.html

base::prop.table(x, margin)

New function ‘proportions()’ and ‘marginSums()’. These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’. for R 4.0.0.

R> m <- matrix(1:4, 2)
R> prop.table(m, 1) # row percentage
          [,1]      [,2]
[1,] 0.2500000 0.7500000
[2,] 0.3333333 0.6666667
R> prop.table(m, 2) # column percentage
          [,1]      [,2]
[1,] 0.3333333 0.4285714
[2,] 0.6666667 0.5714286

stats::xtabs()

stats::ftable()

> ftable(Titanic, row.vars = 1:3)
                   Survived  No Yes
Class Sex    Age                   
1st   Male   Child            0   5
             Adult          118  57
      Female Child            0   1
             Adult            4 140
2nd   Male   Child            0  11
             Adult          154  14
      Female Child            0  13
             Adult           13  80
3rd   Male   Child           35  13
             Adult          387  75
      Female Child           17  14
             Adult           89  76
Crew  Male   Child            0   0
             Adult          670 192
      Female Child            0   0
             Adult            3  20
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
             Survived  No Yes
Class Sex                    
1st   Male            118  62
      Female            4 141
2nd   Male            154  25
      Female           13  93
3rd   Male            422  88
      Female          106  90
Crew  Male            670 192
      Female            3  20
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
             Survived  No Yes
Sex    Class                 
Male   1st            118  62
       2nd            154  25
       3rd            422  88
       Crew           670 192
Female 1st              4 141
       2nd             13  93
       3rd            106  90
       Crew             3  20
> str(Titanic)
 table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
 - attr(*, "dimnames")=List of 4
  ..$ Class   : chr [1:4] "1st" "2nd" "3rd" "Crew"
  ..$ Sex     : chr [1:2] "Male" "Female"
  ..$ Age     : chr [1:2] "Child" "Adult"
  ..$ Survived: chr [1:2] "No" "Yes"
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
> x
          gear  3  4  5
cyl vs am              
4   0  0        0  0  0
       1        0  0  1
    1  0        1  2  0
       1        0  6  1
6   0  0        0  0  0
       1        0  2  1
    1  0        2  2  0
       1        0  0  0
8   0  0       12  0  0
       1        0  0  2
    1  0        0  0  0
       1        0  0  0
> ftable(x, row.vars = c(2, 4))
        cyl  4     6     8   
        am   0  1  0  1  0  1
vs gear                      
0  3         0  0  0  0 12  0
   4         0  0  0  2  0  0
   5         0  1  0  1  0  2
1  3         1  0  2  0  0  0
   4         2  6  2  0  0  0
   5         0  1  0  0  0  0
> 
> ## Start with expressions, use table()'s "dnn" to change labels
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
         dnn = c("Cylinders", "V/S", "Transmission", "Gears"))

          Cylinders     4     6     8   
          Transmission  0  1  0  1  0  1
V/S Gears                               
0   3                   0  0  0  0 12  0
    4                   0  0  0  2  0  0
    5                   0  1  0  1  0  2
1   3                   1  0  2  0  0  0
    4                   2  6  2  0  0  0
    5                   0  1  0  0  0  0

tracemem, data type, copy

How to avoid copying a long vector

Tell if the current R is running in 32-bit or 64-bit mode

8 * .Machine$sizeof.pointer

where sizeof.pointer returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.

32- and 64-bit

See R-admin.html.

  • For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
  • Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
  • Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).

Handling length 2^31 and more in R 3.0.0

From R News for 3.0.0 release:

There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.

In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error

> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector

However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):

> system.time(x <- seq(1,2^31))
   user  system elapsed
  8.604  11.060 120.815
> length(x)
[1] 2147483648
> length(x)/2^20
[1] 2048
> gc()
             used    (Mb) gc trigger    (Mb)   max used    (Mb)
Ncells     183823     9.9     407500    21.8     350000    18.7
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
>

Note:

  1. 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
  2. On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
  3. My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
  4. For large dataset, it makes sense to use database or specially crafted packages like bigmemory or ff or bigstatsr.
  5. [[<- for index 2^31 fails

NA in index

  • Question: what is seq(1, 3)[c(1, 2, NA)]?

Answer: It will reserve the element with NA in indexing and return the value NA for it.

  • Question: What is TRUE & NA?

Answer: NA

  • Question: What is FALSE & NA?

Answer: FALSE

  • Question: c("A", "B", NA) != "" ?

Answer: TRUE TRUE NA

  • Question: which(c("A", "B", NA) != "") ?

Answer: 1 2

  • Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?

Answer: TRUE TRUE FALSE

  • Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?

Answer: TRUE TRUE FALSE

Conclusion: In order to exclude empty or NA for numerical or character data type, we can use which() or a convenience function keep.complete(x) <- function(x) x != "" & !is.na(x). This will guarantee return logical values and not contain NAs.

Don't just use x != "" OR !is.na(x).

Some functions

Constant and 'L'

Add 'L' after a constant. For example,

for(i in 1L:n) { }

if (max.lines > 0L) { }

label <- paste0(n-i+1L, ": ")

n <- length(x);  if(n == 0L) { }

Vector/Arrays

R indexes arrays from 1 like Fortran, not from 0 like C or Python.

remove integer(0)

How to remove integer(0) from a vector?

Append some elements

append() and its after argument

setNames()

Assign names to a vector

z <- setNames(1:3, c("a", "b", "c"))
# OR
z <- 1:3; names(z) <- c("a", "b", "c")
# OR
z <- c("a"=1, "b"=2, "c"=3) # not work if "a", "b", "c" is like x[1], x[2], x[3].

Factor

labels argument

We can specify the factor levels and new labels using the factor() function.

sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))

factor(rev(letters[1:3]), labels = c("A", "B", "C"))
# C B A
# Levels: A B C

Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when()

cut(
     c(0, 10, 30), 
     breaks = c(0, 30, 50, Inf), 
     labels = c("Young", "Middle-aged", "Elderly")
 )  # Default include.lowest = FALSE
# [1] <NA>  Young Young
  • ?cut
    set.seed(1)
    x <- rnorm(100)
    facVar <- cut(x, c(min(x), -1, 1, max(x)), labels = c("low", "medium", "high"))
    table(facVar, useNA = "ifany")
    facVar
    #   low medium   high   <NA> 
    #    10     74     15      1 
    

    Note the option include.lowest = TRUE is needed when we use cut() + quantile(); otherwise the smallest data will become NA since the intervals have the format (a, b].

    x2 <- cut(x, quantile(x, 0:2/2), include.lowest = TRUE) # split x into 2 levels
    x2 <- cut(x, quantile(x, 0:3/3), include.lowest = TRUE) # split x into 3 levels
    
    library(tidyverse); library(magrittr)
    set.seed(1)
    breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
    x <- runif(50)
    bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
    
    data.frame(sc=x, bins=bins) %>% 
      group_by(bins) %>% 
      summarise(n=n()) %>% 
      ggplot(aes(x = bins, y = n)) + 
        geom_col(color = "black", fill = "#90AACB") + 
        theme_minimal() + 
        theme(axis.text.x = element_text(angle = 90)) + 
        theme(legend.position = "none") + coord_flip()
    
  • A Guide to Using the cut() Function in R
  • tibble object
    library(tidyverse)
    tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
           age_cat = case_when(
              age_yrs < 2 ~ "baby",
              age_yrs < 13 ~ "kid",
              age_yrs < 20 ~ "teen",
              TRUE         ~ "adult")
    )
    
  • R tip: Learn dplyr’s case_when() function
    case_when(
      condition1 ~ value1, 
      condition2 ~ value2,
      TRUE ~ ValueAnythingElse
    )
    # Example
    case_when(
      x %%2 == 0 ~ "even",
      x %%2 == 1 ~ "odd",
      TRUE ~ "Neither even or odd"
    )
    

How to change one of the level to NA

https://stackoverflow.com/a/25354985. Note that the factor level is removed.

x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA

Creating missing values in factors

Concatenating two factor vectors

Not trivial. How to concatenate factors, without them being converted to integer level?.

unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))

droplevels()

droplevels(): drop unused levels from a factor or, more commonly, from factors in a data frame.

factor(x , levels = ...) vs levels(x) <-

Note levels(x) is to set/rename levels, not reorder. Use relevel() or factor() to reorder.

levels()
plyr::revalue()
forcats::fct_recode()
rename levels
factor(, levels) reorder levels
sizes <- factor(c("small", "large", "large", "small", "medium"))
sizes
#> [1] small  large  large  small  medium
#> Levels: large medium small

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

sizes3 <- sizes
levels(sizes3) <- c("small", "medium", "large") # rename, not reorder
                                                # large -> small
                                                # medium -> medium
                                                # small -> large 
sizes3
# [1] large  small  small  large  medium
# Levels: small medium large

A regression example.

set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
 
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x2           0.99620    0.09462   10.53   <2e-16 ***

# Wrong way when we want to change the baseline level to '2'
# No change on the model fitting except the apparent change on the variable name in the printout
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x1           0.99620    0.09462   10.53   <2e-16 ***

# Correct way if we want to change the baseline level to '2'
# The estimate was changed by flipping the sign from the original data
sample_data$x <- relevel(x, ref = "2")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  1.96425    0.06770   29.01   <2e-16 ***
# x1          -0.99620    0.09462  -10.53   <2e-16 ***

stats::relevel()

relevel. This function can only be used to change the reference level of a factor variable. It does not directly create an arbitrary order of levels. That is, it is useful in lm() or aov(), etc.

reorder(), levels() and boxplot()

  • How to Reorder Boxplots in R: A Comprehensive Guide (tapply() method, simple & effective)
  • reorder().This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
    # Syntax:
    # newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
    
    (bymedian <- with(InsectSprays, reorder(spray, count, median)) )
    class(bymedian)
    levels(bymedian)
    boxplot(count ~ bymedian, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # boxplots are sorted according to the new levels
    boxplot(count ~ spray, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # not sorted
    
  • Statistics Sunday: My 2019 Reading (reorder function)

factor() vs ordered()

factor(levels=c("a", "b", "c"), ordered=TRUE)
# ordered(0)
# Levels: a < b < c

factor(levels=c("a", "b", "c"))
# factor(0)
# Levels: a b c

ordered(levels=c("a", "b", "c"))
# Error in factor(x, ..., ordered = TRUE) : 
#  argument "x" is missing, with no default

Data frame

stringsAsFactors = FALSE

http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/

We can use options(stringsAsFactors=FALSE) forces R to import character data as character objects.

In R 4.0.0, stringAsFactors=FALSE will be default. This also affects read.table() function.

check.names = FALSE

Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.

> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
  1a 2a
1  1  1
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1   1   1
2   2   2

Create unique rownames: make.unique()

groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
rownames(mydf) <- make.unique(groupCodes)

data.frame() will change rownames

class(df2)
# [1] "matrix" "array"
rownames(df2)[c(9109, 44999)]
# [1] "A1CF"     "A1BG-AS1"
rownames(data.frame(df2))[c(9109, 44999)]
# [1] "A1CF"     "A1BG.AS1"

Print a data frame without rownames

# Method 1. 
rownames(df1) <- NULL

# Method 2. 
print(df1, row.names = FALSE)

Convert data frame factor columns to characters

Convert data.frame columns from factors to characters

# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)

# Method 2:
bob[] <- lapply(bob, as.character)

To replace only factor columns:

# Method 1:
i <- sapply(bob, is.factor)
bob[i] <- lapply(bob[i], as.character)

# Method 2:
library(dplyr)
bob %>% mutate_if(is.factor, as.character) -> bob

Sort Or Order A Data Frame

How To Sort Or Order A Data Frame In R

  1. df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
  2. library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
  3. library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
  4. library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)

data.frame to vector

df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))

class(df)
# [1] "data.frame"
class(t(df))
# [1] "matrix" "array"
class(unlist(df))
# [1] "numeric"

# Method 1: Convert data frame to matrix using as.matrix()
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)   # [1] 1 2 3 4 5 6
vec2 <- c(mat)

# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))

# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6

# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3 
#  1  2  3  4  5  6 
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6

Q: Why as.vector(df) cannot convert a data frame into a vector?

A: The as.vector function cannot be used directly on a data frame to convert it into a vector because a data frame is a list of vectors (i.e., its columns) and as.vector only removes the attributes of an object to create a vector. When you apply as.vector to a data frame, R does not know how to concatenate these independent columns (which could be of different types) into a single vector. Therefore, it doesn’t perform the operation. Therefore as.vector() returns the underlying list structure of the data frame instead of converting it into a vector.

However, when you transpose the data frame using t(), it gets converted into a matrix. A matrix in R is a vector with dimensions. Therefore, all elements of the matrix must be of the same type. If they are not, R will coerce them to be so. Once you have a matrix, as.vector() can easily convert it into a vector because all elements are of the same type.

Using cbind() to merge vectors together?

It’s a common mistake to try and create a data frame by cbind()ing vectors together. This doesn’t work because cbind() will create a matrix unless one of the arguments is already a data frame. Instead use data.frame() directly. See Advanced R -> Data structures chapter.

cbind NULL and data.frame

cbind can't combine NULL with dataframe. Add as.matrix() will fix the problem.

merge

Special character in the matched variable can create a trouble when we use merge() or dplyr::inner_join(). I guess R internally turns df2 (a matrix but not a data frame) to a data frame (so rownames are changed if they contain special character like "-"). This still does not explain the situation when I

class(df1); class(df2)
# [1] "data.frame"  # 2 x 2
# [1] "matrix" "array" # 52439 x 2
rownames(df1)
# [1] "A1CF"     "A1BG-AS1"
merge(df1, df2[c(9109, 44999), ], by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BG-AS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1      A1CF     0    0    3.01147   1.189578
rownames(df2)[38958:38959]
# [1] "ITFG2-AS1"  "ADGRD1-AS1"

rownames(df1)[2] <- "A1BGAS1"
rownames(df2)[44999] <- "A1BGAS1"
merge(df1, df2, by= 0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1   A1BGAS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578

is.matrix: data.frame is not necessarily a matrix

See ?matrix. is.matrix returns TRUE if x is a vector and has a "dim" attribute of length 2 and FALSE otherwise.

An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by

X <- data.frame(x=1:2, y=3:4)

The 'X' object is NOT a vector and it does NOT have the "dim" attribute. It has only 3 attributes: "names", "row.names" & "class". Note that dim() function works fine and returns correctly though there is not "dim" attribute.

Another example that is a data frame but not a matrix is the built-in object cars; see ?matrix. It is not a vector

Convert a data frame to a matrix: as.matrix() vs data.matrix()

If I have a data frame X which recorded the time of some files.

  • is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
  • as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
  • data.matrix(X) will convert the time to numerical values. So use data.matrix() if the data is numeric. The returned object is not a data frame anymore.
# latex directory contains cache files from knitting an rmarkdown file
X <- list.files("latex/", full.names = T) %>%
     grep("RData", ., value=T) %>% 
     file.info() %>%  
     `[`("mtime")
X %>% is.data.frame() # TRUE
X %>% is.matrix() # FALSE
X %>% as.matrix() %>% is.matrix() # TRUE
X %>% data.matrix() %>% is.matrix() # TRUE
X %>% as.matrix() %>% "["(1:2, ) # timestamps
X %>% data.matrix() %>% "["(1:2, ) # numeric
  • The as.matrix() function is used to coerce an object into a matrix. It can be used with various types of R objects, such as vectors, data frames, and arrays.
  • The data.matrix() function is specifically designed for converting a data frame into a matrix by coercing all columns to numeric values. If the data frame contains non-numeric columns, such as character or factor columns, data.matrix() will convert them to numeric values if possible (e.g., by converting factors to their integer codes).
  • See the following example where as.matrix() and data.matrix() return different resuls.
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
mat <- as.matrix(df)
mat
#      a   b  
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array" 
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array" 
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"

matrix vs data.frame

Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.

Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
 [1] "time"   "status" "treat"  "AKT1"   "BRAF"   "FLOT2"  "MTOR"   "PCK2"   "PIK3CA"
[10] "RAF1"  
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10

Case 2:

ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])   # OK

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

The length of a matrix and a data frame is different.

> length(matrix(1:6, 3, 2))
[1] 6
> length(data.frame(matrix(1:6, 3, 2)))
[1] 2
> x[1]
  X1
1  1
2  2
3  3
4  4
5  5
6  6
> x1
[1] 1 2 3 4 5 6

So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.

How to Remove Duplicates

How to Remove Duplicates in R with Example

Convert a matrix (not data frame) of characters to numeric

Just change the mode of the object

tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
     a     b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)     # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"

> mode(tmp) <- "numeric"
> sum(tmp)
[1] 1.917

Convert Data Frame Row to Vector

as.numeric() or c()

Convert characters to integers

mode(x) <- "integer"

Non-Standard Evaluation

Understanding Non-Standard Evaluation. Part 1: The Basics

Select Data Frame Columns in R

This is part of series of DATA MANIPULATION IN R from datanovia.com

  • pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
  • select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
  • select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
  • Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names

Another way is to the dollar sign $ operator (?"$") to extract rows or column from a data frame.

class(USArrests)  # "data.frame"
USArrests$"Assault"

Note that for both data frame and matrix objects, we need to use the [ operator to extract columns and/or rows.

USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#         Murder Assault
# Alabama   13.2     236
# Alaska    10.0     263
USArrests[c("Murder", "Assault")]  # all rows

tmp <- data(package="datasets")
class(tmp$results)  # "matrix" "array" 
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix

Note for a data.table object, we can extract columns using the column names without double quotes.

data.table(USArrests)[1:2, list(Murder, Assault)]

Add columns to a data frame

How to add columns to a data frame in R

Exclude/drop/remove data frame columns

# method 1
df = subset(mydata, select = -c(x,z) )

# method 2
drop <- c("x","z")
df = mydata[,!(names(mydata) %in% drop)]

# method 3: dplyr
mydata2 = select(mydata, -a, -x, -y)
mydata2 = select(mydata, -c(a, x, y))
mydata2 = select(mydata, -a:-y)
mydata2 = mydata[,!grepl("^INC",names(mydata))]

Remove Rows from the data frame

Remove Rows from the data frame in R

Danger of selecting rows from a data frame

> dim(cars)
[1] 50  2
> data.frame(a=cars[1,], b=cars[2, ])
  a.speed a.dist b.speed b.dist
1       4      2       4     10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a  b
speed 4  4
dist  2 10

Creating data frame using structure() function

Creating data frame using structure() function in R

Create an empty data.frame

https://stackoverflow.com/questions/10689055/create-an-empty-data-frame

# the column types default as logical per vector(), but are then overridden
a = data.frame(matrix(vector(), 5, 3,
               dimnames=list(c(), c("Date", "File", "User"))),
               stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)

# similar to above
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))

# different data type
a <- data.frame(x1 = character(),
                x2 = numeric(),
                x3 = factor(),
                stringsAsFactors = FALSE)

Objects from subsetting a row in a data frame vs matrix

  • Subsetting creates repeated rows. This will create unexpected rownames.
    R> z <- data.frame(x=1:3, y=2:4)
    R> rownames(z) <- letters[1:3]
    R> rownames(z)[c(1,1)]
    [1] "a" "a"
    R> rownames(z[c(1,1),])
    [1] "a"   "a.1"
    R> z[c(1,1), ]
        x y
    a   1 2
    a.1 1 2
    
  • Convert a dataframe to a vector (by rows) The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
    str(trainData)
    # 'data.frame':	503 obs. of  500 variables:
    #  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
    #  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
    # ...
    trainData[1:3, 1:3]
    #        bm001      bm002    bm003
    # 1  0.4289449 0.05676296 1.657966
    # 2  1.0000000 1.00000000 1.000000
    # 3 -0.5000000 0.50000000 0.500000
    o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
    # Warning message:
    # In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
    #   row names were found from a short variable and have been discarded
    

    'trees' data from the 'datasets' package

    trees[1:3,]
    #   Girth Height Volume
    # 1   8.3     70   10.3
    # 2   8.6     65   10.3
    # 3   8.8     63   10.2
    
    # Wrong ways:
    data.frame(trees[1,] , trees[2,])
    #   Girth Height Volume Girth.1 Height.1 Volume.1
    # 1   8.3     70   10.3     8.6       65     10.3
    data.frame(time=trees[1,] , status=trees[2,])
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=as.vector(trees[1,]) , status=as.vector(trees[2,]))
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=c(trees[1,]) , status=c(trees[2,]))
    # time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    
    # Right ways:
    # method 1: dropping row names
    data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,]))) 
    # OR
    data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
    #   time status
    # 1  8.3    8.6
    # 2 70.0   65.0
    # 3 10.3   10.3
    # method 2: keeping row names
    data.frame(time=t(trees[1,]) , status=t(trees[2,]))
    #          X1   X2
    # Girth   8.3  8.6
    # Height 70.0 65.0
    # Volume 10.3 10.3
    data.frame(time=unlist(trees[1,]) , status=unlist(trees[2,]))
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    # Method 3: convert a data frame to a matrix
    is.matrix(trees)
    # [1] FALSE
    trees2 <- as.matrix(trees)
    data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    dim(trees[1,])
    # [1] 1 3
    dim(trees2[1, ])
    # NULL
    trees[1, ]  # notice the row name '1' on the left hand side
    #   Girth Height Volume
    # 1   8.3     70   10.3
    trees2[1, ]
    #  Girth Height Volume
    #    8.3   70.0   10.3
    

Convert a list to data frame

How to Convert a List to a Data Frame in R.

# method 1
data.frame(t(sapply(my_list,c)))

# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)

# method 3
library(data.table)
rbindlist(my_list)

tibble and data.table

Clean a dataset

How to clean the datasets in R

matrix

Define and subset a matrix

  • Matrix in R
    • It is clear when a vector becomes a matrix the data is transformed column-wisely (byrow = FALSE, by default).
    • When subsetting a matrix, it follows the format: X[rows, colums] or X[y-axis, x-axis].
data <- c(2, 4, 7, 5, 10, 1)
A <- matrix(data, ncol = 3)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7   10
# [2,]    4    5    1

A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]    7   10

Prevent automatic conversion of single column to vector

use drop = FALSE such as mat[, 1, drop = FALSE].

complete.cases(): remove rows with missing in any column

It works on a sequence of vectors, matrices and data frames.

NROW vs nrow

?nrow. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.

matrix (column-major order) multiply a vector

> matrix(1:6, 3,2)
     [,1] [,2]
[1,]    1    4
[2,]    2    5
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3) # c(1,2,3) will be recycled to form a matrix. Good quiz.
     [,1] [,2]
[1,]    1    4
[2,]    4   10
[3,]    9   18
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
     [,1] [,2]
[1,]    1   16
[2,]    4    5
[3,]    9   12

add a vector to all rows of a matrix

add a vector to all rows of a matrix. sweep() or rep() is the best.

sparse matrix

R convert matrix or data frame to sparseMatrix

To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, as.vector().

Attributes

Names

Useful functions for dealing with object names. (Un)Setting object names: stats::setNames(), unname() and rlang::set_names()

Print a vector by suppressing names

Use unname. sapply(, , USE.NAMES = FALSE).

format.pval/print p-values/format p values

format.pval(). By default it will show 5 significant digits (getOption("digits")-2).

> format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"     
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"

R> pvalue
[1] 0.0004632104
R> print(pvalue, digits =20)
[1] 0.00046321036188223807528
R> format.pval(pvalue)
[1] "0.00046321"
R> format.pval(pvalue * 1e-1)
[1] "4.6321e-05"
R> format.pval(0.00004632)
[1] "4.632e-05"
R> getOption("digits")
[1] 7

Customize R: options()

Change the default R repository, my .Rprofile

Change R repository

Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local .Rprofile settings take precedence.

For example, I can specify the R mirror I like by creating a single line .Rprofile file under my home directory. Another good choice of repository is cloud.r-project.org.

Type file.edit("~/.Rprofile")

local({
  r = getOption("repos")
  r["CRAN"] = "https://cran.rstudio.com/"
  options(repos = r)
})
options(continue = "  ", editor = "nano")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
  .Last <- function() try(savehistory("~/.Rhistory"))
}

Change the default web browser for utils::browseURL()

When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put

options(browser='seamonkey')

in the .Rprofile of your home directory. If the browser is not in the global PATH, we need to put the full path above.

For one-time only purpose, we can use the browser option in help.start() function:

> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...

We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See

Change the default editor

On my Linux and mac, the default editor is "vi". To change it to "nano",

options(editor = "nano")

Change prompt and remove '+' sign

See https://stackoverflow.com/a/1448823.

options(prompt="R> ", continue=" ")

digits

  • signif() rounds x to n significant digits.
    R> signif(pi, 3)
    [1] 3.14
    R> signif(pi, 5)
    [1] 3.1416
    
  • The default digits 7 may be too small. For example, if a number is very large, then we may not be able to see (enough) value after the decimal point. The acceptable range is 1-22. See the following examples

In R,

> options()$digits # Default
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
[1] 100000.1
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11

In Python,

>>> 100000.07 + .04
100000.11

Disable scientific notation in printing: options(scipen)

How to Turn Off Scientific Notation in R?

This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.

> numer = 29707; denom = 93874
> c(numer/denom, numer, denom) 
[1] 3.164561e-01 2.970700e+04 9.387400e+04

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

# Method 2. Change the global option
> options(scipen=999)
> numer/denom
[1] 0.3164561
> c(numer/denom, numer, denom)
[1]     0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]     0.8 29707.0 93874.0

Suppress warnings: options() and capture.output()

Use options(). If warn is negative all warnings are ignored. If warn is zero (the default) warnings are stored until the top--level function returns.

op <- options("warn")
options(warn = -1)
....
options(op)

# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)

suppressWarnings()

suppressWarnings( foo() )

foo <- capture.output( 
 bar <- suppressWarnings( 
 {print( "hello, world" ); 
   warning("unwanted" )} ) ) 

capture.output()

str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")

Converts warnings into errors

options(warn=2)

demo() function

  • How to wait for a keypress in R? PS readline() is different from readLines().
    for(i in 1:2) { print(i); readline("Press [enter] to continue")}
    
  • Hit 'ESC' or Ctrl+c to skip the prompt "Hit <Return> to see next plot:"
  • demo() uses options() to ask users to hit Enter on each plot
    op <- options(device.ask.default = ask)  # ask = TRUE
    on.exit(options(op), add = TRUE)
    

sprintf

paste, paste0, sprintf

this post, 3 R functions that I enjoy

sep vs collapse in paste()

  • sep is used if we supply multiple input objects to paste()
  • collapse is used to make the output of length 1. It is commonly used if we have only 1 input object
R> paste("a", "A", sep=",")
[1] "a,A"
R> paste("a", "A", sep=",", collapse="-")
[1] "a,A"
R> paste(c("a", "A"), collapse="-")
[1] "a-A"

R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], collapse = "-")
[1] "a-b-c"

Format number as fixed width, with leading zeros

# sprintf()
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# formatC()
paste("name", formatC(a, width=3, flag="0"), sep="_")
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# gsub()
paste0("bm", gsub(" ", "0", format(5:15)))
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"

formatC and prettyNum (prettifying numbers)

R> (x <- 1.2345 * 10 ^ (-8:4))
 [1] 1.2345e-08 1.2345e-07 1.2345e-06 1.2345e-05 1.2345e-04 1.2345e-03
 [7] 1.2345e-02 1.2345e-01 1.2345e+00 1.2345e+01 1.2345e+02 1.2345e+03
[13] 1.2345e+04
R> formatC(x)
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "0.0001234" "0.001234"
 [7] "0.01235"   "0.1235"    "1.234"     "12.34"     "123.4"     "1234"
[13] "1.234e+04"
R> formatC(x, digits=3)
 [1] "1.23e-08" "1.23e-07" "1.23e-06" "1.23e-05" "0.000123" "0.00123"
 [7] "0.0123"   "0.123"    "1.23"     "12.3"     " 123"     "1.23e+03"
[13] "1.23e+04"
R> formatC(x, digits=3, format="e")
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "1.234e-04" "1.234e-03"
 [7] "1.235e-02" "1.235e-01" "1.234e+00" "1.234e+01" "1.234e+02" "1.234e+03"
[13] "1.234e+04"

R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"

Format(x, scientific = TRUE)

Print numeric data in exponential format, so .0001 prints as 1e-4

Creating publication quality graphs in R

HDF5 : Hierarchical Data Format

HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.

Formats for writing/saving and sharing data

Efficiently Saving and Sharing Data in R

Write unix format files on Windows and vice versa

https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html

with() and within() functions

closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)

mk <- within(mariokart, {
             closePr <- totalPr - shipPr
     })
head(mk) # new column closePr

mk <- mariokart
aggregate(. ~ wheels + cond, mk, mean)
# create mean according to each level of (wheels, cond)

aggregate(totalPr ~ wheels + cond, mk, mean)

tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)

stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals

Plot histograms as lines

https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.

x2=invisible(hist(out2$EB))
y2=invisible(hist(out2$Bench))
z2=invisible(hist(out2$EB0.001))

plot(x=x2$mids, y=x2$density, type="l")
lines(y2$mids, y2$density, lty=2, pwd=2)
lines(z2$mids, z2$density, lty=3, pwd=2)

Histogram with density line

hist(x, prob = TRUE)
lines(density(x), col = 4, lwd = 2)

The overlayed density may looks strange in cases for example counts from single-cell RNASeq or p-values from RNASeq (there is a peak around x=0).

Graphical Parameters, Axes and Text, Combining Plots

statmethods.net

15 Questions All R Users Have About Plots

See 15 Questions All R Users Have About Plots. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.

  1. How To Draw An Empty R Plot? plot.new()
  2. How To Set The Axis Labels And Title Of The R Plots?
  3. How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
  4. How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). ?par.
  5. How To Add Or Change The R Plot’s Legend? legend()
  6. How To Draw A Grid In Your R Plot? grid()
  7. How To Draw A Plot With A PNG As Background? rasterImage() from the png package
  8. How To Adjust The Size Of Points In An R Plot? cex argument
  9. How To Fit A Smooth Curve To Your R Data? loess() and lines()
  10. How To Add Error Bars In An R Plot? arrows()
  11. How To Save A Plot As An Image On Disc
  12. How To Plot Two R Plots Next To Each Other? par(mfrow)[which means Multiple Figures (use ROW-wise)], gridBase package, lattice package
  13. How To Plot Multiple Lines Or Points? plot(), lines()
  14. How To Fix The Aspect Ratio For Your R Plots? asp parameter
  15. What Is The Function Of hjust And vjust In ggplot2?

jitter function

Jitterbox.png

Scatterplot with the "rug" function

require(stats)  # both 'density' and its default method
with(faithful, {
    plot(density(eruptions, bw = 0.15))
    rug(eruptions)
    rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
})

File:RugFunction.png

See also the stripchart() function which produces one dimensional scatter plots (or dot plots) of the given data.

Identify/Locate Points in a Scatter Plot

  • ?identify
  • Using the identify function in R
    plot(x, y)
    identify(x, y, labels = names, plot = TRUE) 
    # Use left clicks to select points we want to identify and "esc" to stop the process
    # This will put the labels on the plot and also return the indices of points
    # [1] 143
    names[143]
    

Draw a single plot with two different y-axes

Draw Color Palette

Default palette before R 4.0

palette() # black, red, green3, blue, cyan, magenta, yellow, gray

# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
palette(tropical)
plot(1:5, 1:5, col=1:5, pch=16, cex=5)

New palette in R 4.0.0

R 4.0: 3 new features, R 4.0.0 now available, and a look back at R's history. For example, we can select "ggplot2" palette to make the base graphics charts that match the color scheme of ggplot2.

R> palette() 
[1] "black"   "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "gray62"
R> palette.pals()
 [1] "R3"              "R4"              "ggplot2"        
 [4] "Okabe-Ito"       "Accent"          "Dark 2"         
 [7] "Paired"          "Pastel 1"        "Pastel 2"       
[10] "Set 1"           "Set 2"           "Set 3"          
[13] "Tableau 10"      "Classic Tableau" "Polychrome 36"  
[16] "Alphabet"
R> palette.colors(palette='R4') # same as palette()
[1] "#000000" "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "#9E9E9E"
R> palette("R3")  # nothing return on screen but palette has changed
R> palette() 
[1] "black"   "red"     "green3"  "blue"    "cyan"    "magenta" "yellow" 
[8] "gray"  
R> palette("R4") # reset to the default color palette; OR palette("default")

R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
R> for(id in palette.pals()) { 
     scales::show_col(palette.colors(palette = id))
     title(id)
     readline("Press [enter] to continue") 
   } 

The palette function can also be used to change the color palette. See Setting up Color Palettes in R

palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
   # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))

cc <- palette()
palette(c(cc,"purple","brown")) # Add two colors
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502

evoPalette

Evolve new colour palettes in R with evoPalette

rtist

rtist: Use the palettes of famous artists in your own visualizations.

SVG

Embed svg in html

svglite

svglite is better R's svg(). It was used by ggsave(). svglite 1.2.0, R Graphics Cookbook.

pdf -> svg

Using Inkscape. See this post.

svg -> png

SVG to PNG using the gyro package

read.table

clipboard

source("clipboard")
read.table("clipboard")

inline text

mydf <- read.table(header=T, text='
 cond yval
    A 2
    B 2.5
    C 1.6
')

http(s) connection

temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt", 
                           ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)

read only specific columns

Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.

x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)), 
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
# 
system.time(x <- read.delim("Methylation450k.txt", 
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))

To know the number of columns, we might want to read the first row first.

library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()

Another method is to use pipe(), cut or awk. See ways to read only selected columns from a file into R

check.names = FALSE in read.table()

gx <- read.table(file, header = T, row.names =1)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH."        "IGHV1.69"    "IGKV1.5"     "IGKV2.24"    "KRTAP13.2"  
# [7] "KRTAP19.1"   "KRTAP2.4"    "KRTAP5.9"    "KRTAP6.3"    "Kua.UEV"  

gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH@"        "IGHV1-69"    "IGKV1-5"     "IGKV2-24"    "KRTAP13-2"  
# [7] "KRTAP19-1"   "KRTAP2-4"    "KRTAP5-9"    "KRTAP6-3"    "Kua-UEV"  

setNames()

Change the colnames. See an example from tidymodels

Testing for valid variable names

Testing for valid variable names

make.names(): Make syntactically valid names out of character vectors

  • make.names()
  • A valid variable name consists of letters, numbers and the dot or underline characters. The variable name starts with a letter or the dot not followed by a number. See R variables.
make.names("abc-d") # [1] "abc.d"

Serialization

If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the post on R mailing list.

> a <- list(1,2,3)
> a_serial <- serialize(a, NULL)
> a_length <- length(a_serial)
> a_length
[1] 70
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)

In C++ process, I receive one int variable first to get the length, and then read <length> bytes from the connection.

socketConnection

See ?socketconnection.

Simple example

from the socketConnection's manual.

Open one R session

con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
close(con1)

Open another R session (client)

con2 <- socketConnection(Sys.info()["nodename"], port = 22131)
# as non-blocking, may need to loop for input
readLines(con2)
while(isIncomplete(con2)) {
   Sys.sleep(1)
   z <- readLines(con2)
   if(length(z)) print(z)
}
close(con2)

Use nc in client

The client does not have to be the R. We can use telnet, nc, etc. See the post here. For example, on the client machine, we can issue

nc localhost 22131   [ENTER]

Then the client will wait and show anything written from the server machine. The connection from nc will be terminated once close(con1) is given.

If I use the command

nc -v -w 2 localhost -z 22130-22135

then the connection will be established for a short time which means the cursor on the server machine will be returned. If we issue the above nc command again on the client machine it will show the connection to the port 22131 is refused. PS. "-w" switch denotes the number of seconds of the timeout for connects and final net reads.

Some post I don't have a chance to read. http://digitheadslabnotebook.blogspot.com/2010/09/how-to-send-http-put-request-from-r.html

Use curl command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

curl --trace-ascii debugdump.txt http://localhost:8080/

Then go to the server,

while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")

close(con1) # return cursor in the client machine

Use telnet command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

sudo apt-get install telnet
telnet localhost 8080
abcdefg
hijklmn
qestst

Go to the server,

readLines(con1, 1)
readLines(con1, 1)
readLines(con1, 1)
close(con1) # return cursor in the client machine

Some tutorial about using telnet on http request. And this is a summary of using telnet.

Subsetting

Subset assignment of R Language Definition and Manipulation of functions.

The result of the command x[3:5] <- 13:15 is as if the following had been executed

`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)

Avoid Coercing Indices To Doubles

1 or 1L

Careful on NA value

See the example below. base::subset() or dplyr::filter() can remove NA subsets.

R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3   3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6

Implicit looping

set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]     # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7

modelling

update()

Extract all variable names in lm(), glm(), ...

all.vars(formula(Model)[-2])

as.formula(): use a string in formula in lm(), glm(), ...

? as.formula
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")

# Method 1. The 'Call' portion of the model is reported as “formula = f” 
# our modeling effort, 
# fully parameterized!
f <- as.formula(
  paste(outcome, 
        paste(variables, collapse = " + "), 
        sep = " ~ "))
print(f)
# mpg ~ cyl + disp + hp + carb

model <- lm(f, data = mtcars)
print(model)

# Call:
#   lm(formula = f, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Method 2. eval() + bquote() + ".()"
format(terms(model))  #  or model$terms
# [1] "mpg ~ cyl + disp + hp + carb"

# The new line of code
model <- eval(bquote(   lm(.(f), data = mtcars)   ))

print(model)
# Call:
#   lm(formula = mpg ~ cyl + disp + hp + carb, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Note if we skip ".()" operator
> eval(bquote(   lm(f, data = mtcars)   ))

Call:
lm(formula = f, data = mtcars)

Coefficients:
(Intercept)          cyl         disp           hp         carb  
  34.021595    -1.048523    -0.026906     0.009349    -0.926863 

reformulate

Simplifying Model Formulas with the R Function ‘reformulate()’

I() function

I() means isolates. See What does the capital letter "I" in R linear regression formula mean?, In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)

Aggregating results from linear model

https://stats.stackexchange.com/a/6862

Replacement function "fun(x) <- a"

What are Replacement Functions in R?

R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
     x[x > value] <- Inf
     x
 }
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]   1   3 Inf Inf

R> "cutoff<-"(x = xx, value = 65)
[1]   1   3 Inf Inf

The statement fun(x) <- a and R will read x <- "fun<-"(x,a)

S3 and S4 methods and signature

Debug an S4 function

  • showMethods('FUNCTION')
  • getMethod('FUNCTION', 'SIGNATURE')
  • debug(, signature)
> args(debug)
function (fun, text = "", condition = NULL, signature = NULL) 

> library(genefilter) # Bioconductor
> showMethods("nsFilter")
Function: nsFilter (package genefilter)
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")

library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code

See the source code of normalizationFactors<- (setReplaceMethod() is used) and the source code of estimateSizeFactors(). We can see how avgTxLength was used in estimateNormFactors().

Another example

library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)

showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"

debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz, 
#      normalization = ssgsea.norm, verbose = verbose, 
#      BPPARAM = BPPARAM))

isdebugged("gsva")
# [1] TRUE
undebug(gsva)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir

class(ir)
## [1] "IRanges"
## attr(,"package")
## [1] "IRanges"

getClassDef(class(ir))
## Class "IRanges" [package "IRanges"]
## 
## Slots:
##                                                                       
## Name:            start           width           NAMES     elementType
## Class:         integer         integer characterORNULL       character
##                                       
## Name:  elementMetadata        metadata
## Class: DataTableORNULL            list
## 
## Extends: 
## Class "Ranges", directly
## Class "IntegerList", by class "Ranges", distance 2
## Class "RangesORmissing", by class "Ranges", distance 2
## Class "AtomicList", by class "Ranges", distance 3
## Class "List", by class "Ranges", distance 4
## Class "Vector", by class "Ranges", distance 5
## Class "Annotated", by class "Ranges", distance 6
## 
## Known Subclasses: "NormalIRanges"

Check if a function is an S4 method

isS4(foo)

How to access the slots of an S4 object

  • @ will let you access the slots of an S4 object.
  • Note that often the best way to do this is to not access the slot directly but rather through an accessor function (e.g. coefs() rather than digging out the coefficients with $ or @). However, often such functions do not exist so you have to access the slots directly. This will mean that your code breaks if the internal implementation changes, however.
  • R - S4 Classes and Methods Hansen. getClass() or getClassDef().

setReplaceMethod()

See what methods work on an object

see what methods work on an object, e.g. a GRanges object:

methods(class="GRanges")

Or if you have an object, x:

methods(class=class(x))

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

?":::"

  • pkg::name returns the value of the exported variable name in namespace pkg
  • pkg:::name returns the value of the internal variable name
base::"+"
stats:::coef.default

predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR  
getS3method("predict", "ppr")

getS3method("t", "test")

methods() + getAnywhere() functions

Read the source code (include Fortran/C, S3 and S4 methods)

S3 method is overwritten

For example, the select() method from dplyr is overwritten by grpreg package.

An easy solution is to load grpreg before loading dplyr.

mcols() and DataFrame() from Bioc S4Vectors package

  • mcols: Get or set the metadata columns.
  • colData: SummarizedExperiment instances from GenomicRanges
  • DataFrame: The DataFrame class extends the DataTable virtual class and supports the storage of any type of object (with length and [ methods) as columns.

For example, in Shrinkage of logarithmic fold changes vignette of the DESeq2paper package

> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
   baseMean   baseVar   allZero dispGeneEst    dispFit dispersion  dispIter dispOutlier   dispMAP
  <numeric> <numeric> <logical>   <numeric>  <numeric>  <numeric> <numeric>   <logical> <numeric>
1  163.5750  8904.607     FALSE  0.06263141 0.03862798  0.0577712         7       FALSE 0.0577712
2  175.3883 59643.515     FALSE  2.25306109 0.03807917  2.2530611        12        TRUE 1.6011440
  Intercept strain_DBA.2J_vs_C57BL.6J SE_Intercept SE_strain_DBA.2J_vs_C57BL.6J WaldStatistic_Intercept
  <numeric>                 <numeric>    <numeric>                    <numeric>               <numeric>
1  6.210188                  1.735829    0.1229354                    0.1636645               50.515872
2  6.234880                  1.823173    0.6870629                    0.9481865                9.074686
  WaldStatistic_strain_DBA.2J_vs_C57BL.6J WaldPvalue_Intercept WaldPvalue_strain_DBA.2J_vs_C57BL.6J
                                <numeric>            <numeric>                            <numeric>
1                                10.60602         0.000000e+00                         2.793908e-26
2                                 1.92280         1.140054e-19                         5.450522e-02
   betaConv  betaIter  deviance  maxCooks
  <logical> <numeric> <numeric> <numeric>
1      TRUE         3  210.4045 0.2648753
2      TRUE         9  243.7455 0.3248949

Pipe

Packages take advantage of pipes

  • rstatix: Pipe-Friendly Framework for Basic Statistical Tests

findInterval()

Related functions are cuts() and split(). See also

Assign operator

  • Earlier versions of R used underscore (_) as an assignment operator.
  • Assignments with the = Operator
  • In R 1.8.0 (2003), the assign operator has been removed. See NEWS.
  • In R 1.9.0 (2004), "_" is allowed in valid names. See NEWS.
R162.png

Operator precedence

The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.

order(), rank() and sort()

If we want to find the indices of the first 25 genes with the smallest p-values, we can use order(pval)[1:25].

> x = sample(10)
> x
 [1]  4  3 10  7  5  8  6  1  9  2
> order(x)
 [1]  8 10  2  1  5  7  4  6  9  3
> rank(x)
 [1]  4  3 10  7  5  8  6  1  9  2
> rank(10*x)
 [1]  4  3 10  7  5  8  6  1  9  2

> x[order(x)]
 [1]  1  2  3  4  5  6  7  8  9 10
> sort(x)
 [1]  1  2  3  4  5  6  7  8  9 10

OS-dependent results on sorting string vector

Gene symbol case.

# mac: 
order(c("DC-UbP", "DC2")) # c(1,2)

# linux: 
order(c("DC-UbP", "DC2")) # c(2,1)

Affymetric id case.

# mac:
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"

# linux
order(c("202800_at", "2028_s_at")) # [1] 1 2
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"

It does not matter if we include factor() on the character vector.

The difference is related to locale. See

# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2

# Or setting the locale to "C"
Sys.setlocale("LC_ALL", "C"); sort(c("DC-UbP", "DC2"))
# Or
Sys.setlocale("LC_COLLATE", "C"); sort(c("DC-UbP", "DC2"))
# But not
Sys.setlocale("LC_ALL", "en_US.UTF-8"); sort(c("DC-UbP", "DC2"))

unique()

It seems it does not sort. ?unique.

# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"

do.call

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

The do.call() function in R: Unlocking Efficiency and Flexibility

Below are some examples from the help.

  • Usage
do.call(what, args, quote = FALSE, envir = parent.frame())
# what: either a function or a non-empty character string naming the function to be called.
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# quote: a logical value indicating whether to quote the arguments.
# envir: an environment within which to evaluate the call. This will be most useful
#        if what is a character string and the arguments are symbols or quoted expressions.
  • do.call() is similar to lapply() but not the same. It seems do.call() can make a simple function vectorized.
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
> lapply(list(imag = 1:3), complex)
$imag
[1] 0+0i
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
> do.call(function(x) x+1, list(1:3))
[1] 2 3 4
  • Applying do.call with Multiple Arguments
> do.call("sum", list(c(1,2,3,NA), na.rm = TRUE))
[1] 6
> do.call("sum", list(c(1,2,3,NA) ))
[1] NA
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> length(tmp)
[1] 3
> tmp[1:4,]
  Var1 Var2 Var3
1    a    1    +
2    b    1    +
3    a    2    +
4    b    2    +
> c(tmp, sep = "")
$Var1
 [1] a b a b a b a b a b a b
Levels: a b

$Var2
 [1] 1 1 2 2 3 3 1 1 2 2 3 3

$Var3
 [1] + + + + + + - - - - - -
Levels: + -

$sep
[1] ""
> do.call("paste", c(tmp, sep = ""))
 [1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[12] "b3-"
  • environment and quote arguments.
> A <- 2
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A)   
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env)  
[1] 4
> do.call(f, list(A), envir = env)   
[1] 2                       # Why?

> eval(call("f", A))                      
[1] 2
> eval(call("f", quote(A)))               
[1] 2
> eval(call("f", A), envir = env)         
[1] 4
> eval(call("f", quote(A)), envir = env)  
[1] 100
> foo <- function(a=1, b=2, ...) { 
         list(arg=do.call(c, as.list(match.call())[-1])) 
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a 
1 
> foo(a=1, b=2, c=3)
$arg
a b c 
1 2 3 
  • do.call() + switch(). See an example from Seurat::NormalizeData.
do.call(
   what = switch(
     EXPR = margin,
     '1' = 'rbind',
     '2' = 'cbind',
     stop("'margin' must be 1 or 2")
   ),
   args = normalized.data
)
switch('a', 'a' = rnorm(3), 'b'=rnorm(4)) # switch returns a value
do.call(switch('a', 'a' = 'rnorm', 'b'='rexp'), args=list(n=4)) # switch returns a function
  • The function we want to call is a string that may change: glmnet
# Suppose we want to call cv.glmnet or cv.coxnet or cv.lognet or cv.elnet .... depending on the case
fun = paste("cv", subclass, sep = ".")
cvstuff = do.call(fun, list(predmat,y,type.measure,weights,foldid,grouped))

expand.grid, mapply, vapply

A faster way to generate combinations for mapply and vapply

do.call vs mapply

  • do.call() is doing what mapply() does but do.call() uses a list instead of multiple arguments. So do.call() more close to base::Map() function.
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
      Var1 
 [1,] "a1+"
 [2,] "b1+"
 [3,] "a2+"
 [4,] "b2+"
 [5,] "a3+"
 [6,] "b3+"
 [7,] "a1-"
 [8,] "b1-"
 [9,] "a2-"
[10,] "b2-"
[11,] "a3-"
[12,] "b3-"
# It does not work if we do not explicitly specify the arguments in mapply()
> mapply(paste, tmp, sep = "")
      Var1 Var2 Var3
 [1,] "a"  "1"  "+" 
 [2,] "b"  "1"  "+" 
 [3,] "a"  "2"  "+" 
 [4,] "b"  "2"  "+" 
 [5,] "a"  "3"  "+" 
 [6,] "b"  "3"  "+" 
 [7,] "a"  "1"  "-" 
 [8,] "b"  "1"  "-" 
 [9,] "a"  "2"  "-" 
[10,] "b"  "2"  "-" 
[11,] "a"  "3"  "-" 
[12,] "b"  "3"  "-" 
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
# [1] 1.326108 9.885284
set.seed(1)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
dim(x) # [1]  2 1000
rowMeans(x)
# [1]  1.032209 10.104131
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))

do.call vs lapply

What's the difference between lapply and do.call? It seems to me the best usage is combining both functions: do.call(..., lapply())

  • lapply returns a list of the same length as X, each element of which is the result of applying FUN to the corresponding element of X.
  • do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it. It is widely used, for example, to assemble lists into simpler structures (often with rbind or cbind).
  • Map applies a function to the corresponding elements of given vectors... Map is a simple wrapper to mapply which does not attempt to simplify the result, similar to Common Lisp's mapcar (with arguments being recycled, however). Future versions may allow some control of the result type.
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
[1] "numeric"

$Sepal.Width
[1] "numeric"

$Petal.Length
[1] "numeric"

$Petal.Width
[1] "numeric"

$Species
[1] "factor"

> x <- lapply(iris, class)
> do.call(c, x)
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species 
   "numeric"    "numeric"    "numeric"    "numeric"     "factor" 

https://stackoverflow.com/a/10801902

  • lapply applies a function over a list. So there will be several function calls.
  • do.call calls a function with a list of arguments (... argument) such as c() or rbind()/cbind() or sum or order or "[" or paste. So there is only one function call.
> X <- list(1:3,4:6,7:9)
> lapply(X,mean)
1
[1] 2

2
[1] 5

3
[1] 8
> do.call(sum, X)
[1] 45
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
[1] 45
> do.call(mean, X) # Error
> do.call(rbind,X)
     [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9
> lapply(X,rbind)
1
     [,1] [,2] [,3]
[1,]    1    2    3

2
     [,1] [,2] [,3]
[1,]    4    5    6

3
     [,1] [,2] [,3]
[1,]    7    8    9
> mapply(mean, X, trim=c(0,0.5,0.1))
[1] 2 5 8
> mapply(mean, X) 
[1] 2 5 8

Below is a good example to show the difference of lapply() and do.call() - Generating Random Strings.

> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0)
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0, collapse= "")
1
[1] "YDGA"

2
[1] "BWKN"

> do.call(paste0, x)
[1] "YB" "DW" "GK" "AN"

do.call + rbind + lapply

Lots of examples. See for example this one for creating a data frame from a vector.

x <- readLines(textConnection("---CLUSTER 1 ---
 3
 4
 5
 6
 ---CLUSTER 2 ---
 9
 10
 8
 11"))

 # create a list of where the 'clusters' are
 clust <- c(grep("CLUSTER", x), length(x) + 1L)

 # get size of each cluster
 clustSize <- diff(clust) - 1L

 # get cluster number
 clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])

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

 result

     Object Cluster
[1,] "3"    "1"
[2,] "4"    "1"
[3,] "5"    "1"
[4,] "6"    "1"
[5,] "9"    "2"
[6,] "10"   "2"
[7,] "8"    "2"
[8,] "11"   "2"

A 2nd example is to sort a data frame by using do.call(order, list()).

Another example is to reproduce aggregate(). aggregate() = do.call() + by().

attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# the above approach give the same result as the following
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)

Run examples

When we call help(FUN), it shows the document in the browser. The browser will show

example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed

How to get examples from help file, example()

Code examples in the R package manuals:

# How to run all examples from a man page
example(within)

# How to check your examples?
devtools::run_examples() 
testthat::test_examples()

See this post. Method 1:

example(acf, give.lines=TRUE)

Method 2:

Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)

"[" and "[[" with the sapply() function

Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.

sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)

is the same as

sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])

Dealing with dates

  • Find difference
# Convert the dates to Date objects
date1 <- as.Date("6/29/21", format="%m/%d/%y")
date2 <- as.Date("11/9/21", format="%m/%d/%y")

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

# OR using the lubridate package
library(lubridate)
# Convert the dates to Date objects
date1 <- mdy("6/29/21")
date2 <- mdy("11/9/21")
interval(date1, date2) %/% months(1)
d1 = date()
class(d1) # "character"
d2 = Sys.Date()
class(d2) # "Date"

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

library(lubridate); ymd("20140108") # "2014-01-08 UTC"
mdy("08/04/2013") # "2013-08-04 UTC"
dmy("03-04-2013") # "2013-04-03 UTC"
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland") 
# "2011-08-03 10:15:03 NZST"
?Sys.timezone
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues

Nonstandard/non-standard evaluation, deparse/substitute and scoping

f <- function(x) {
  substitute(x)
}
f(1:10)
# 1:10
class(f(1:10)) # or mode()
# [1] "call"
g <- function(x) deparse(substitute(x))
g(1:10)
# [1] "1:10"
class(g(1:10)) # or mode()
# [1] "character"
  • quote(expr) - similar to substitute() but do nothing?? noquote - print character strings without quotes
mode(quote(1:10))
# [1] "call"
  • eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))

subset1 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR

subset2 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌 
subset2(sample_df, a == condition) # 👍
  • deparse(expr) - turns unevaluated expressions into character strings. For example,
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", " 
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                    
[4] "NULL"     

> deparse(args(lm), width=20)
[1] "function (formula, data, "        "    subset, weights, "           
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, "   
[5] "    y = FALSE, qr = TRUE, "       "    singular.ok = TRUE, "        
[7] "    contrasts = NULL, "           "    offset, ...) "               
[9] "NULL"

Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).

f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3

f1(1:3)
f2(1:3)
f3(1:3)

# Or
myfun <- function(f, a) {
    eval(parse(text = f))(a)
}
myfun("f1", 1:3)
myfun("f2", 1:3)
myfun("f3", 1:3)

# Or with lapply
method <- c("f1", "f2", "f3")
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method

library() accept both quoted and unquoted strings

How can library() accept both quoted and unquoted strings. The key lines are

  if (!character.only) 
     package <- as.character(substitute(package))

Lexical scoping

The ‘…’ argument

Functions

Function argument

Argument matching from R Language Definition manual.

Argument matching is augmented by the functions

Access to the partial matching algorithm used by R is via pmatch.

Check function arguments

Checking the inputs of your R functions: match.arg() , stopifnot()

stopifnot(): function argument sanity check

  • stopifnot(). stopifnot is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.
    stopifnot(condition1, condition2, ...)
    
  • Mining R 4.0.0 Changelog for Nuggets of Gold

Lazy evaluation in R functions arguments

R function arguments are lazy — they’re only evaluated if they’re actually used.

  • Example 1. By default, R function arguments are lazy.
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
  • Example 2. If you want to ensure that an argument is evaluated you can use force().
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
  • Example 3. Default arguments are evaluated inside the function.
f <- function(x = ls()) {
  a <- 1
  x
}

# ls() evaluated inside f:
f()
# [1] "a" "x"

# ls() evaluated in global environment:
f(ls())
# [1] "add"    "adders" "f" 
  • Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
x <- NULL
if (!is.null(x) && x > 0) {

}

Use of functions as arguments

Just Quickly: The unexpected use of functions as arguments

body()

Remove top axis title base plot

Return functions in R

anonymous function

In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a lambda function is defined without a name, while a regular function is defined with a name.

  • See Tidyverse page
  • But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, lapply(list(1,2,3), function(x) { x * x })
  • you can use lambda functions with many other functions in R that take a function as an argument. Some examples include sapply, apply, vapply, mapply, Map, Reduce, Filter, and Find. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
    Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
    
  • purrr anonymous function
  • The new pipe and anonymous function syntax in R 4.1.0
  • Functional programming from Advanced R
  • What are anonymous functions in R.
    > (function(x) x * x)(3)
    [1] 9
    > (\(x) x * x)(3)
    [1] 9

Backtick sign, infix/prefix/postfix operators

The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in Advanced R and What do backticks do in R?.

iris %>%  `[[`("Species")

infix operator.

1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix

Use with functions like sapply, e.g. sapply(1:5, `+`, 3) .

Error handling and exceptions, tryCatch(), stop(), warning() and message()

  • http://adv-r.had.co.nz/Exceptions-Debugging.html
  • Catch Me If You Can: Exception Handling in R
  • Temporarily disable warning messages
    # Method1: 
    suppressWarnings(expr)
    
    # Method 2:
    <pre>
    defaultW <- getOption("warn") 
    options(warn = -1) 
    [YOUR CODE] 
    options(warn = defaultW)
    
  • try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
    out <- try({
      a <- 1
      b <- "x"
      a + b
    })
    
    elements <- list(1:10, c(-1, 10), c(T, F), letters)
    results <- lapply(elements, log)
    is.error <- function(x) inherits(x, "try-error")
    succeeded <- !sapply(results, is.error)
    
  • tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
    tryCatch(expr, ..., finally)
    
    show_condition <- function(code) {
      tryCatch(code,
        error = function(c) "error",
        warning = function(c) "warning",
        message = function(c) "message"
      )
    }
    show_condition(stop("!"))
    #> [1] "error"
    show_condition(warning("?!"))
    #> [1] "warning"
    show_condition(message("?"))
    #> [1] "message"
    show_condition(10)
    #> [1] 10
    

    Below is another snippet from available.packages() function,

    z <- tryCatch(download.file(....), error = identity)
    if (!inherits(z, "error")) STATEMENTS
    
  • The return class from tryCatch() may not be fixed.
    result <- tryCatch({
      # Code that might generate an error or warning
      log(99)
    }, warning = function(w) {
      # Code to handle warnings
      print(paste("Warning:", w))
    }, error = function(e) {
      # Code to handle errors
      print(paste("Error:", e))
    }, finally = {
      # Code to always run, regardless of whether an error or warning occurred
      print("Finished")
    })   
    # character type. But if we remove 'finally', it will be numeric.
    
  • Capture message, warnings and errors from a R function

suppressMessages()

suppressMessages(expression)

List data type

Create an empty list

out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)

outlist <- as.list(seq(nfolds))

Using $ in R on a List

How to Use Dollar sign in R

Calling a function given a list of arguments

> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5

Descend recursively through lists

x[[c(5,3)]] is the same as x[[5]][[3]]. See ?Extract.

Avoid if-else or switch

?plot.stepfun.

y0 <- c(1,2,4,3)
sfun0  <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)

tt <- seq(0, 3, by = 0.1)
op <- par(mfrow = c(2,2))
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
     main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")

for(i in 1:3)
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)

par(op)

File:StepfunExample.svg

Open a new Window device

X11() or dev.new()

par()

?par

text size (cex) and font size on main, lab & axis

Examples (default is 1 for each of them):

  • cex.main=0.9
  • cex.sub
  • cex.lab=0.8, font.lab=2 (x/y axis labels)
  • cex.axis=0.8, font.axis=2 (axis/tick text/labels)
  • col.axis="grey50"

An quick example to increase font size (cex.lab, cex.axis, cex.main) and line width (lwd) in a line plot and cex & lwd in the legend.

plot(x=x$mids, y=x$density, type="l", 
     xlab="p-value", ylab="Density", lwd=2, 
     cex.lab=1.5, cex.axis=1.5, 
     cex.main=1.5, main = "")
lines(y$mids, y$density, lty=2, pwd=2)
lines(z$mids, z$density, lty=3, pwd=2)
legend('topright',legend = c('Method A','Method B','Method C'),
       lty=c(2,1,3), lwd=c(2,2,2), cex = 1.5, xjust = 0.5, yjust = 0.5)

ggplot2 case (default font size is 11 points):

  • plot.title
  • plot.subtitle
  • axis.title.x, axis.title.y: (x/y axis labels)
  • axis.text.x & axis.text.y: (axis/tick text/labels)
ggplot(df, aes(x, y)) +
  geom_point() +
  labs(title = "Title", subtitle = "Subtitle", x = "X-axis", y = "Y-axis") +
  theme(plot.title = element_text(size = 20),
        plot.subtitle = element_text(size = 15),
        axis.title.x = element_text(size = 15),
        axis.title.y = element_text(size = 15),
        axis.text.x = element_text(size = 10),
        axis.text.y = element_text(size = 10))

Default font

layout

reset the settings

op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1) 
....
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1

mtext (margin text) vs title

mgp (axis tick label locations or axis title)

  1. The margin line (in ‘mex’ units) for the axis title, axis labels and axis line. Note that ‘mgp[1]’ affects the axis ‘title’ whereas ‘mgp[2:3]’ affect tick mark labels. The default is ‘c(3, 1, 0)’. If we like to make the axis labels closer to an axis, we can use mgp=c(1.5, .5, 0) for example.
    • the default is c(3,1,0) which specify the margin line for the axis title, axis labels and axis line.
    • the axis title is drawn in the fourth line of the margin starting from the plot region, the axis labels are drawn in the second line and the axis line itself is the first line.
  2. Setting graph margins in R using the par() function and lots of cow milk
  3. Move Axis Label Closer to Plot in Base R (2 Examples)
  4. http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the labels/axis title (i.e. xlab and ylab in plot), the second the tick-mark labels, and third the tick marks. The default is c(3, 1, 0).

move axis title closer to axis

title(ylab="Within-cluster variance", line=0, 
      cex.lab=1.2, family="Calibri Light")

pch and point shapes

File:R pch.png

See here.

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

lty (line type)

File:R lty.png

Line types in R: Ultimate Guide For R Baseplot and ggplot

See here.

ggpubr::show_line_types()

las (label style)

0: The default, parallel to the axis

1: Always horizontal

2: Perpendicular to the axis

3: Always vertical

oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots

no.readonly

R语言里par(no.readonly=TURE)括号里面这个参数什么意思?, R-par()

Non-standard fonts in postscript and pdf graphics

https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41


NULL, NA, NaN, Inf

https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/

save()/load() vs saveRDS()/readRDS() vs dput()/dget() vs dump()/source()

  1. saveRDS() can only save one R object while save() does not have this constraint.
  2. saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See this post.
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)

dput: Writes an ASCII text representation of an R object. The object name is not written (unlike dump).

$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato", 
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos", 
"ast", "trig", "platelet", "protime", "stage")

> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5, 
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L, 
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")

User 'verbose = TRUE' in load()

When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.

What are RDS files anyways

Archive Existing RDS Files

==, all.equal(), identical()

  • ==: exact match
  • all.equal: compare R objects x and y testing ‘near equality’
  • identical: The safe and reliable way to test two objects for being exactly equal.
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE

Be careful about using "==" to return an index of matches in the case of data with missing values.

R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1

See also the testhat package.

I found a case when I compare two objects where 1 is generated in Linux and the other is generated in macOS that identical() gives FALSE but all.equal() returns TRUE. The difference has a magnitude only e-17.

waldo

diffobj: Compare/Diff R Objects

https://cran.r-project.org/web/packages/diffobj/index.html

testthat

tinytest

tinytest: Lightweight but Feature Complete Unit Testing Framework

ttdo adds support of the 'diffobj' package for 'diff'-style comparison of R objects.

Numerical Pitfall

Numerical pitfalls in computing variance

.1 - .3/3
## [1] 0.00000000000000001388

Sys.getpid()

This can be used to monitor R process memory usage or stop the R process. See this post.

Sys.getenv() & make the script more portable

Replace all the secrets from the script and replace them with Sys.getenv("secretname"). You can save the secrets in an .Renviron file next to the script in the same project.

$ for v in 1 2; do MY=$v Rscript -e "Sys.getenv('MY')"; done
[1] "1"
[1] "2"
$ echo $MY
2

How to write R codes

  • Code smells and feels from R Consortium
    • write simple conditions,
    • handle class properly,
    • return and exit early,
    • polymorphism,
    • switch() [e.g., switch(var, value1=out1, value2=out2, value3=out3). Several examples in glmnet ]
    • case_when(),
    • %||%.
  • 5 Tips for Writing Clean R Code – Leave Your Code Reviewer Commentless
    • Comments
    • Strings
    • Loops
    • Code Sharing
    • Good Programming Practices

How to debug an R code

Debug R

Locale bug (grep did not handle UTF-8 properly PR#16264)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264

Path length in dir.create() (PR#17206)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)

install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths()

R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"

On Mac & R 3.4.0 (it's fine)

> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"

On Linux & R 3.3.1 (ARM)

> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"

On Linux & R 3.4.1 (*Problematic*)

> Sys.getenv("R_LIBS_USER")
[1] ""
> .libPaths()
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"

I need to specify the lib parameter when I use the install.packages command.

> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
Error in library(devtools) : there is no package called 'devtools'

# Specify lib.loc parameter will not help with the dependency package
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
Error: package or namespace load failed for 'devtools':
 .onLoad failed in loadNamespace() for 'devtools', details:
  call: loadNamespace(name)
  error: there is no package called 'withr'

# A solution is to redefine .libPaths
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works

A better solution is to specify R_LIBS_USER in ~/.Renviron file or ~/.bash_profile; see ?Startup.

Using external data from within another package

https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/

How to run R scripts from the command line

https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line

How to exit a sourced R script

Decimal point & decimal comma

Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark

setting seed locally (not globally) in R

https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r

R's internal C API

https://github.com/hadley/r-internals

cleancall package for C resource cleanup

Resource Cleanup in C and the R API

Random number generator

#include <R.h>

void myunif(){
  GetRNGstate();
  double u = unif_rand();
  PutRNGstate();
  Rprintf("%f\n",u);
}
$ R CMD SHLIB r_rand.c
$ R
R> dyn.load("r_rand.so")
R> set.seed(1)
R> .C("myunif")
0.265509
list()
R> .C("myunif")
0.372124
list()
R> set.seed(1)
R> .C("myunif")
0.265509
list()

Test For Randomness

Different results in Mac and Linux

Random numbers: multivariate normal

Why MASS::mvrnorm() gives different result on Mac and Linux/Windows?

The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See

rle() running length encoding

citation()

citation()
citation("MASS")
toBibtex(citation())

R not responding request to interrupt stop process

R not responding request to interrupt stop process. R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts. It seems to match with the case I'm running (dist() function).

Monitor memory usage

  • x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
  • Windows: memory.size(max=TRUE)
  • Linux
    • RStudio: htop -p PID where PID is the process ID of /usr/lib/rstudio/bin/rsession, not /usr/lib/rstudio/bin/rstudio. This is obtained by running x <- rnorm(2*1e8). The object size can be obtained through print(object.size(x), units = "auto"). Note that 1e8*8/2^20 = 762.9395.
    • R: htop -p PID where PID is the process ID of /usr/lib/R/bin/exec/R. Alternatively, use htop -p `pgrep -f /usr/lib/R/bin/exec/R`
    • To find the peak memory usage grep VmPeak /proc/$PID/status
  • mem_used() function from pryr package. It is not correct or useful if I use it to check the value compared to the memory returned by jobload in biowulf. So I cannot use it to see the memory used in running mclapply().
  • peakRAM: Monitor the Total and Peak RAM Used by an Expression or Function
  • Error: protect () : protection stack overflow and ?Memory

References:

Monitor Data

Monitoring Data in R with the lumberjack Package

Pushover

Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}

pushoverr

Resource

Books

  • Efficient R programming by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the Appendix. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
    # R 3.4.1
    .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
    setwd("/tmp/efficientR/")
    bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
    # generated pdf file is located _book/_main.pdf
    
    bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
    # generated epub file is located _book/_main.epub.
    # This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
    # but it is OK to run in an R terminal
    

Videos

“Do More with R” video tutorials. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.

Webinar

useR!

R consortium

https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured

Blogs, Tips, Socials, Communities

Bug Tracking System

https://bugs.r-project.org/bugzilla3/ and Search existing bug reports. Remember to select 'All' in the Status drop-down list.

Use sessionInfo().

License

Some Notes on GNU Licenses in R Packages

Why Dash uses the mit license (and not a copyleft gpl license)

Interview questions

  • Does R store matrices in column-major order or row-major order?
    • Matrices are stored in column-major order, which means that elements are arranged and accessed by columns. This is in contrast to languages like Python, where matrices (or arrays) are typically stored in row-major order.
  • Explain the difference between == and === in R. Provide an example to illustrate their use.
    • The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.
  • What is the purpose of the apply() function in R? How does it differ from the for loop?
    • The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.
  • Describe the concept of factors in R. How are they used in data manipulation and analysis?
    • Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.
  • What is the significance
of the attach() and detach() functions in R? When should they be used?
    • A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.
  • Explain the concept of vectorization in R. How does it impact the performance of R code?
    • Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
  • Describe the difference between data.frame and matrix in R. When would you use one over the other?
    • A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
    • A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
    • You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.
  • What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
    • The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
    • It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
    • To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.