Brief R Tutorial

June 6, 2008

The best way to go through this tutorial is to first install a version of R (see installation section below)
and type the commands along with the examples given. This way you can see for yourself what output each
command gives.

Contents
1 Introduction 2

2 R Basics - Installation, Starting, Quitting, and Objects 2

3 Entering or Reading Data Into R 2

3.1 Typing Data In Manually . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.2 Reading Data From a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.3 Accessing Elements of Data Arrays, Vectors, or Matrices . . . . . . . . . . . . . . . . . . . . . 4
3.4 Determining Sizes of Data Structures/Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

4 Basic R Commands You Should Know 4

5 Statistical Commands You Should Know 4

6 Writing Your Own Functions and Sourcing Code 5

6.1 Writing a Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
6.2 For and While Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
6.3 Logical Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
6.4 Sourcing Code and Setting the Working Directory . . . . . . . . . . . . . . . . . . . . . . . . 6

7 Graphics 7
7.1 Creating a pdf or ps File of an R Figure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

8 Writing Output To A File 7

9 Installing Packages 8

1
1 Introduction
R is a statistical programming language that provides many built in functions for performing statistical
analysis. R is also flexible enough to allow users to write their own functions and source code written in a
text editor such as NotePad or Emacs. The advantage to learning R is that R is very easy to learn and easy
to use. However, R is quite slow when doing heavy computational work (such as Bayesian algorithms) and
so using R for heavy computing is not recommended.
This tutorial provides a very brief introduction to how to use R. This document will go through some of
the most commonly used R functions but will in no way cover all of the functions in R. To learn advanced
R functions, you should use Internet searches with the key word CRAN which stands for Comprehensive
R Archive Network. Using the key word CRAN in addition to other key words about the function you are
looking for will generally produce a lot of results.
If you forget the specific syntax for the R functions listed in this tutorial, you can simply type ?function
and R will return the documentation for function which will provide almost all the necessary information
for using function. As an example, typing ?qnorm will return the documentation for the R function qnorm.
Alternatively you can use help(functionname) which does the same thing that ?functionname does.

2 R Basics - Installation, Starting, Quitting, and Objects

R can be downloaded and installed for free from the website http://cran.r-project.org. This web page
provides detailed instructions for installing R on any operating system. If you are using the department
computers, you do not need to install R as it has already been installed for you.
Accessing R is different for every operating system. For windows users, simply double click the R icon
that is created after installation. For Mac users, you can double click the R icon under your Applications
menu. On Linux and the department computers, in the terminal window (if you don’t know what a terminal
window is then please read the Linux tutorial) simply type R at the command prompt and R will be opened
within the terminal window.
When you open R, no matter the operating system you are using, you will see the command prompt
symbol “>” which simply means that R is waiting for you to give it a command. To quit R, simply type “q()”
in the command prompt and R will ask you if you want to save the workspace before quitting. Generally,
it is NOT a good idea to save workspace because if you continually save your work space future programs
you run may become effected by previously run programs. Nevertheless, it is up to you whether or not you
want to save your workspace. After answering this question by typing “y” or “n” R will quit.
R is an object oriented language so output produced from running functions will be output to the screen
unless saved as an object. You create an object in R by using the “<-” command. For example, the code

x < − rnorm(10,10,10)

will save the output from using the rnorm() function into x. Having saved the output from rnorm() as
the object x, you can print the output from the rnorm() function by simply typing x into the command
prompt.

3 Entering or Reading Data Into R

Before you can begin programming you will need to enter or read data into R. This can be done in several
ways. You can either type the data in manually (generally not a good idea) as a matrix, vector, or array or
you can read in your data from a file (much better idea). This section will cover how to do both as you will
use both at some point.

2
3.1 Typing Data In Manually
Here are some commands if you want to create or type in data manually:
• c() - short for “concatenate”; creates an array with a single row of the elements of its arguments. For
example, x <- c(1,2,3,4) will create an array object x which contains the numbers 1, 2, 3, 4.

• matrix(data,nrow=,ncol=) - takes the information in data and creates a matrix with nrow rows and
ncol columns. For example, the code matrix(c(1,2,3,4),nrow=2,ncol=2) creates a 2×2 matrix with 1
and 2 in the first column and 3 and 4 in the second column. You can create a matrix by rows using
matrix(c(1,2,3,4),nrow=2,ncol=2,byrow=T) where the rows will now be (1,2) and (3,4) respectively.

• array(data,dim=c(a,b)) - create from data an array with a rows and b columns. For example ar-
ray(c(1,2,3,4),dim=c(2,2)) will create a 2×2 array. You can create a higher dimensional array by
giving more arguments to the dim argument.
• seq(from,to,by=a) - creates a single row array by creating a sequence from from to to by by. For
example, seq(0,1,by=.001) creates a sequence of numbers from 0 to 1 stepping in increments of 0.001.

• diag() - create a diagonal matrix of a single row array. For example, diag(c(1,1,1)) creates the 3×3
identity matrix.
• cbind(a,b) - creates a matrix with the vector a as its first column and b as its second column.
• rbind(a,b) - creates a matrix with the vector a as its first row and b as its second row.

• rep(data,times) - creates a single row array by repeating the numbers in data, times many times. For
example, rep(2,3) will create an array by repeating the number 2, 3 times.

3.2 Reading Data From a File

Oftentimes instead of typing in data sets manually, you will want to read data in from a text or other type
of file. To do so you can use any of the options below depending on your situation:

• read.table(filename,header=,sep=””) - will read the file filename as a data frame (matrix) which is delim-
ited by the character specified in sep=””. For example, x <- read.table(“mydata.txt”,header=T,sep=“,”)
will read the data in mydata.txt which is a comma separated file and has a header (header just means
that you have the variable names at the top of the file).
• scan(filename) - this command will do the same as read.table() however scan() will read in the entire
file as a single variable. Bottom line, if you have more than one variable in a data set you need to use
read.table() because scan() will not distinguish between variables in the file.
• data(dataname) - R has several built in data sets. The data() function will simply load the built in
data set. The data will automatically be saved as the object dataname.
• attach() - When you read in data using read.table(,header=T,) or data() and save the data as an object
you won’t be able to directly access the variables names. For example, say you read in a file mydata.txt
which has two variables var1 and var2 using the code x <- read.table(“mydata.txt”,header=T,sep=” ”).
Then if you type var1 into the R command prompt it will say that the variable var1 is not found.
Instead, the object x has two variable names associated with it, namely var1 and var2. You can access
the numbers of var1 by typing x$var1. However, if you type attach(x) and then type var1, the computer
will now recognize the variable var1.

3
3.3 Accessing Elements of Data Arrays, Vectors, or Matrices
Once you have either read data in from a file or created data manually by using any of the above commands,
you often need to access a specific element or variable within the object. Here is a brief tutorial about how
to access elements of data frames, matrices, and arrays.

• Let x be a single row array. To access the ith element of x type x[i].
• Let x be an n × n matrix. To access the ij th element type x[i,j].
• Let x be a data frame or any other object with individual variables var1 and var2 contained within it.
To access the values saved under the variable name var1 type x$var1. Alternatively, you can attach x
and type var1.
• You can see if any object x has any variable names associated with it by typing names(x).

3.4 Determining Sizes of Data Structures/Objects

Often you will need to know the dimension of R objects. For example, you may want to know how many
observations are contained in an array. Use any of the following functions to see the dimensions an object.
• length() - returns the length of an object. If the object is an matrix or an array with more than 1
dimension, it will return the total number of elements within that object.

• dim() - returns the dimension of an object. For example, dim(mymatrix) will return the dimension of
mymatrix. If mymatrix is a single dimensional array, it will return NULL and you should use length()
instead.

4 Basic R Commands You Should Know

Here are list of very basic functions you will use frequently:
• A%*%B - does matrix multiplication of the matrices A and B.
• solve(A) - computes the inverse of the matrix A.

• t(A) - computes the transpose of the matrix A.

• chol(A) - computes the cholesky decomposition of the matrix A.
• eigen(A) - computes the eigenvalues and eigenvectors of the square matrix A.
• apply(A,dim,function) - For a matrix or multi-dimensional array A, apply will perform the function
function on the dimth dimension of A. For example, if A is a p × k matrix, the code apply(A,1,sum) will
sum the rows (the first dimension) of A. Alternatively, the code apply(A,2,sum) will sum the columns
of A.
• kronecker(A,B) - takes the kronecker product of A and B.

5 Statistical Commands You Should Know

• mean(),median(),max(),min(),sd(),var() - computes the mean, median, max, min, standard deviation,
and variance of an object (typically a data array), respectively. If the object is a multidimensional
array, the function var() will assume that the columns of the object are different variables and return
the variance-covariance matrix.

4
 Distribution CDF PDF Inverse CDF Draw Randomly
Normal pnorm() dnorm() qnorm() rnorm()
Binomial pbinom() dbinom() qbinom() rbinom()
Negative Binomial pnbinom() dnbinom() qnbinom() rnbinom()
Poisson ppois() dpois() qpois() rpois()
Beta pbeta() dbeta() qbeta() rbeta()
χ2 pchisq() dchisq() qchisq() rchisq()
Exponential pexp() dexp() qexp() rexp()
F pf() df() qf() rf()
Gamma pgamma() dgamma() qgamma() rgamma()
t pt() dt() qt() rt()
Uniform punif() dunif() qunif() runif()

Table 1: Table of distribution and their corresponding R functions. Use the help() command to see their
exact syntax.

• lm(),glm() - calculate a linear model and generalized linear model for a data set. You will learn A LOT
more about these functions in first year courses so no detail is given here. If you want more detail use
the help command.
• summary() - summarizes the information of an object. For example if x is a lm object then summary(lm)
will summarize the fit of the linear model. Once again, you will learn a great deal more about this
function in your first year courses so not a lot of detail is given here.
• Distributions - Table 1 displays the basic functions for the most commonly used distributions. Use
the help command to see the exact syntax for these commands. For example, help(rnorm) will show
you the syntax you need to draw randomly from a normal distribution.

6 Writing Your Own Functions and Sourcing Code

In almost all but the simplest of cases, you will want R to execute a series of commands instead of executing
commands line by line as is R’s default. To tackle bigger programming jobs, R allows the user to write a
series of commands in a separate editor (such as Emacs, TextEdit, or Notepad) and then R will “source”
the code and run the code all together. This section will go through the details of how to write your own
functions and programs in a separate editor and then source the code in R.

6.1 Writing a Function

You can declare your own function in R using the function() command. The syntax is

myfunction() <- function(input1,input2,...){

Function code
}

In the above example, the object myfunction() is now a function which takes inputs input1 and input2,
and runs the code specified in the Function code. As a specific example, consider

mlevar <- function(datavector){

var(datavector)*(n-1)/(n)
}

5
In this case, the function mlevar calculates the MLE of σ 2 instead of the unbiased estimator of σ 2 , s2 .
Now, the code mle <- mlevar(mydata) will assign to the object mle the value equal to the MLE of σ 2 for your
data. If your function is more complex you will need to specify which object calculated in your function to
return to the user using the return() function. For example,

mlevar <- function(datavector){

output <- var(datavector)*(n-1)/(n)
return(output)
}

will return the value in output as the output of the function mlevar.

6.2 For and While Loops

The syntax for writing for loops in R is

for(i in 1:N){
Loop Code
}

where Loop Code should be substituted with the code you want to run in the loop. The syntax for while
loops in R is

while(logical argument){
Loop Code
}

where logical argument should be an argument that is either true or false at each iteration. WARNING
R is VERY slow at doing loops. If you are doing a lot of looping in your code, you are advised to use
MATLAB or C++ as these languages are MUCH faster than using R.

6.3 Logical Arguments

The following table gives a few logical argument operators used in R.

Code Interpretation
< Less than
<= Less than or equal to
&& And
>= Greater than or equal to
> Greater than

6.4 Sourcing Code and Setting the Working Directory

Once you have completed writing your code in a separate text editor such as Emacs, TextEdit, or Notepad,
save the file using the “.R” file extension. For example, “mycode.R”. You can then tell R to “source” your
code and run the entire program. To do so simply type the command,

source(“mycode.R”)

6
and R will run the entire program. If there are bugs in your code (and there inevitably will be) R will
print the error in the R console window for you to return to your code file and make changes.
In the above source example, R will return an error if the file “mycode.R” is not in the current working
directory. For example, if R is working from the directory “./Desktop/” and the file “mycode.R” is located
in “./home/” directory then you need to set the working directory R is working out of. To do so, use the
command setwd(“directorypath”) where directorypath is the folder you want R to work out of. If you are using
unix or a department machine, you can avoid this by simply changing the directory to the desired directory
and then opening R (for more on this see the Unix tutorial).

7 Graphics
One of the main reasons that people choose to use R is the great flexibility that R gives in generating
graphics. While R can generate a wide variety of graphics, the most commonly used graphics functions are
summarized here:
• plot(x,y) - plots the points in x and y on a scatter plot.

• lines(x,y) - plots the points in x and y on the open figure (does not create a new graphic) and connects
the points in x and y with a solid black line.
• points(x,y) - adds the points in x and y on the open figure (does not create a new figure as plot() does).
• title(“string”) - adds the “string” as the title of the figure

• barplot(height,. . . ) - creates a bar plots with heights of the bars equal to the data array height.
• hist(x) - draws a histogram of the data array x
• legend() - adds a legend to the current graphic
You should definitely spend some time messing around with these functions to get practice at generating
plots. A good exercise would be to draw a picture of the standard normal distribution.

7.1 Creating a pdf or ps File of an R Figure

R has commands which allow the user to create a pdf or ps file of a figure created in the R environment.
To do so use either the pdf() or postscript() commands followed by the dev.off() command. For example, the
code:
pdf("./myfigure.pdf")
plot(x,y)
dev.off()

will create a pdf file of the figure generated by the plot(x,y) command. The dev.off() command tells R that
you are done creating the figure and it can now be output the the pdf file.
While you can replace pdf() with postscript(), it is recommended that you create pdf figures because you
will most likely be using pdflatex to compile your LATEXdocument.

8 Writing Output To A File

Having finished running a program, you can then output a data array or matrix to a text or other file type
using the write.table() command. As an example, the code write.table(mydata,file=“filename.dat”) would write
the data in the object mydata to the file “filename.dat.” This command is especially useful for outputting,
say, draws from a posterior distribution drawn using Metropolis-Hastings algorithms.

7
9 Installing Packages
R has many packages available (free of charge) that will expand the basic function package of R to include
more complex functions. For example, R has no default function that draws from a multivariate normal
distribution. However the package mvtnorm has a function rmvnorm() that will draw from a multivariate
normal distribution
To install a package, type install.packages(“packagename”). Once you have typed this, R will prompt
you to choose an installation mirror from which to download the packages. Pick a mirror (it doesn’t matter
which one). R will then automatically install the package for you. Once you install a package you do not
need to install it ever again.
Just installing a package is not enough to access the functions available in a package. You also need to
source the functions in that package by typing library(packagename) and R will load the functions available
in that package for you.
While courses will generally talk to you about what packages to install, the packages that are most
commonly used within the department are:
• The coda package - loads functions for calculating convergence diagnostics
• The mvtnorm package - loads functions to draw from a multivariate normal and multivariate t-
distribution.

• The xtable package - loads functions to output a table generated in R to latex code.
• The R2WinBugs package - loads functions which will output your R code to a WinBugs file for running
a Gibbs sampler. More on this in STA 290.