Mines-Intro-To-R

In order to use R packages from CRAN, you need to set up a personal library in ~/R/library, and tell R to use a proxy to access the web: http://www-proxy.ensmp.fr:8080

This should be no problem if you copy and paste the following R code into your ~/.Rprofile

ldir <- "~/R/library"
if(!file.exists(ldir)){
  dir.create(ldir,recursive=TRUE)
}
.libPaths(c(ldir,.libPaths()))
options(browser="firefox")
options(repos=c("http://cran.univ-lyon1.fr/",
          "http://cran.r-project.org"))
Sys.setenv(http_proxy="http://www-proxy.ensmp.fr:8080")

If you installed R on your personal computer then you probably don't have a proxy, and you probably have a newer version of R that does the user library configuration automatically.

Then open a terminal, type R, and copy/paste the following lines of code to install the required packages

install.packages("ROCR")
source("http://bioconductor.org/biocLite.R")
biocLite("ALL")

The basic idea is that R is a command line interpreter. Write code in a text editor like medit, which is installed by default on the Mines computers and has basic support for syntax highlighting of R code. When you want to execute the R code, you can either copy/paste code into R running in a terminal, or in R do

source("my.script.R")

which will execute all the code in the file my.script.R

To learn the basics of R you should print out a copy of http://cran.r-project.org/doc/contrib/Short-refcard.pdf

• <- and = are both assignment operators
• f(a,b,c) calls function f with arguments a,b,c. do.call(f,list(a,b,c)) does the same thing.
• for(x in L){commands} loops over items in L
• install.packages("kernlab") downloads and installs the kernlab package and all its dependencies from the Comprehensive R Archive Network (CRAN).
• library(kernlab) loads the kernlab package and all its dependencies, which should be already installed on your computer.
• plot(x,y) creates a scatterplot based on the vectors x and y, and par() controls graphical parameters.
• table(x,y) shows a contingency table of all unique values in the vectors x and y.

R contains an extensive help system. To illustrate, say we want to find a function that gives us the Student t-distribution. First type help.search("student") and you will see a list with relevant help pages. The one you want is TDist. Type ?TDist and you will see a help page that explains:

• the density is dt()
• the CDF is pt()
• the inverse CDF (quantile function) is qt()
• random numbers can be generated with rt()

To get help on a function, type ? and then the function name. For example, ?plot shows you the help page for the plot function. Or type help.start() to get a HTML help interface.

R is a vectorized language. That is, when you type x*y it means elementwise multiplication of the vectors x and y.

## DONT DO THIS:
z <- x
for(i in 1:length(x)){
  z[i] <- x[i] * y[i]
}
## YES DO THIS: (does the same but much more efficient)
z <- x * y

At the R command line, you can use TAB completion for variables, functions, and arguments. For example if you forgot the name for the function that can read a data table, you can just type "read", then TAB twice will give you a list of possible functions. After typing "read.t" TAB will fill in and give you "read.table". Add a parenthesis to make it "read.table(" and hitting TAB twice will show you a list of valid keyword arguments to the read.table function.

Some useful keyboard shortcuts. C-a means hold control and type a. M-f means hold Alt and type f.

shortcut	function
C-p or up arrow	recall previous command
C-n or down arrow	recall next command
C-a	go to beginning of line
C-e	go to end of line
M-f	go forward 1 word
M-b	go backward 1 word
M-d	delete word in front
C-delete	delete word in back
C-k	delete until end of line
C-y	yank (paste) deleted word(s)

To use it on the Mines Paristech unix machines, execute the following code in a terminal to download and setup ESS:

cd
mkdir R
cd R
wget http://ess.r-project.org/downloads/ess/ess-5.13.tgz
tar xf ess-5.13.tgz
cat >> ~/.emacs <<EOF
(add-to-list 'load-path "~/R/ess-5.13/lisp")
(load "ess-site")
(setq ess-eval-visibly-p nil)
(setq ess-ask-for-ess-directory nil)
(require 'ess-eldoc)
(show-paren-mode 1)
EOF
wget http://cbio.ensmp.fr/~jvert/svn/tutorials/practical/svmbasic/svmbasic.R
wget http://cbio.ensmp.fr/~jvert/svn/tutorials/practical/svmbasic/general.R

This code should also work on any other Linux system with Emacs and R.

Vincent Goulet maintains excellent Emacs distributions that work out-of-the box with R

http://vgoulet.act.ulaval.ca/emacs/windows

Now you can do open up an R code file in Emacs from the command line using, for example "emacs svmbasic.R &"

Alternatively, from within Emacs you can start R by typing M-x R ENTER.

When writing R code from within Emacs you get context-sensitive help. For example, after typing "read.table(" Emacs will display the list of arguments for the read.table function.

Additionally, you can use the following keyboard commands in a R code file to evaluate R code:

Shortcut	Function
C-c C-n	Send 1 line of R code
C-c SPACE	Mark current position
C-c C-r	Send code between mark and point
C-c C-l	Send entire buffer
C-c TAB or C-c C-i	Complete

Furthermore the R from within Emacs buffer has similar shortcuts to R in the terminal, but with the following exceptions:

Shortcut	Function
M-p	Recall previous command
M-n	Recall next command
C-p	Move cursor up
C-n	Move cursor down

Further information about Emacs is beyond the scope of this guide, but I suggest you take the Emacs tutorial by typing C-h t from within Emacs. Or download the French version and open it in Emacs

http://www.linux-france.org/article/appli/emacs/TUTORIAL.fr