The R Language
1
Dr. Smruti R. Sarangi and Ms. Hameedah Sultan
Computer Science and Engineering
IIT Delhi
2 Overview of R
Language for statistical computing and data analysis
Freely available under GPL v2
Extensive library support
Programming paradigms
procedural
functional
object-oriented
General matrix computation (similar to Matlab)
3 Running R
Command Line
Just type R
The R command prompt comes up
> .....
With a GUI
R Studio
R Commander
4 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
5 Normal Variables
We can use <- as the assignment operator in R
> x <- 4
(set x to 4)
For printing the value of x
>x
[1] 4
OR, > print(x)
[1] 4
6 A Numeric Vector
Simplest data structure
Numeric vector
> v <- c(1,2,3)
<- is the assignment operator
c is the list concatenation operator
To print the value, v
Type : > v
Output: [1] 1 2 3
7 A vector is a full fledged variable
Let us do the following:
> 1/v
[1] 1.0000000 0.5000000 0.3333333
>v+2
[1] 3 4 5
We can treat a vector as a regular variable
For example, we can have:
> v1 <- v / 2
> v1
[1] 0.5 1.0 1.5
8 Creating a vector with vectors
> v <- c (1,2,3)
>v
[1] 1 2 3
> vnew <- c (v,0,v)
> vnew
[1] 1 2 3 0 1 2 3
The c operator concatenates all the vectors
9 Functions on Vectors and Complex
Numbers
If v is a vector
Here, are a few of the functions that take vectors as
inputs:
mean(v), max(v), sqrt(v), length(v), sum(v), prod(v),
sort (v) (in ascending order)
> x <- 1 + 1i
> y <- 1i
>x*y
[1] -1+1i
10 Generating Vectors
Suppose we want a vector of the form:
(1,2,3,... 100)
We do not have to generate it manually.
We can use the following commands:
> v <- 1:100
OR
> v <- seq(1,100)
seq takes an additional argument, which is the difference
between consecutive numbers:
seq (1,100,10) gives (1,11,21,31 ... , 91)
rep (2,5) generates a vector (2, 2, 2, 2, 2)
11 Boolean Variables and Vectors
R recognizes the constants: TRUE, FALSE
TRUE corresponds to 1
FALSE corresponds to 0
We can define a vector of the form:
v <- c (TRUE, FALSE, TRUE)
We can also define a logical vector
Can be created with logical operators: <, <=, >=,
==, !=, & and I
> v <- 1:9 > 5
> v
[1] FALSE FALSE FALSE FALSE FALSE TRUE
TRUE TRUE TRUE
12 String Vectors
Similarly, we can have a vector of strings
> vec <- c (“f1”, “f2”, “f3”)
> vec
[1] "f1" "f2" "f3“
The paste function can be used to create a vector of
strings
paste(1:3, 3:5,sep="*")
[1] "1*3" "2*4" "3*5"
It takes two vectors of the same length, and an optional
argument, sep. The ith element of the result string,
contains the ith elements of both the arguments,
separated by the string specified by sep.
13 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
14 Factors
Factor Definition: A vector used to specify
a grouping (classification) of objects
in other vectors.
Consider the following problem:
We have a vector of the type of the Nationality of
students, and a vector of their marks in a given
subject.
AIM: Find the average scores per nationality.
15 Graphical View of the Problem
Indian 6
Chinese 8
Indian
Indian 7
Chinese
Chinese 9
Russian
Indian 8
Factor
Russian 10
Nationality Marks
# character starts
16 Code a comment
> nationalities <- c ("Indian", "Chinese", "Indian", "Chinese",
"Indian", "Russian") # create a factor
> marks <- c (6, 8, 7, 9, 8, 10)
> fac <- factor(nationalities)
> fac
[1] Indian Chinese Indian Chinese Indian Russian
Levels: Chinese Indian Russian
The levels of a factor indicate the categories
17 Code - II
Now let us apply the factor to the marks
vector
> results <- tapply (marks, fac, mean)
Works on each compute the mean
factor
element of the list in each category
List of marks
18 Time for the results
> results
Chinese Indian Russian
8.5 7.0 10.0
Let us now apply the sum function
> tapply (marks, fac, sum)
Chinese Indian Russian
17 21 10
19 levels and table
> levels (fac)
[1] "Chinese" "Indian" "Russian"
> table (fac)
fac
Chinese Indian Russian
2 3 1
Let us assume that the factor is fac.
fac is
[1] Indian Chinese Indian Chinese Indian Russian
Levels: Chinese Indian Russian
levels returns a vector containing all the
unique labels
table returns a special kind of array that
20 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
21 Arrays and Matrices
Generic array function
Creates an array. Takes two arguments:
data_vector vector of values
dimension_vector
Example:
> array (1:10, c(2,5))
[,1 [,2] [,3] [,4] [,5]
[1,] 1 3 5 7 9
[2,] 2 4 6 8 10
The numbers are laid out in column major order.
Count from 1, Not 0
22 Other ways to make arrays
Take a vector, and assign it dimensions
> v <- c (1,2,3,4)
> dim(v) <- c(2,2)
>v
[,1] [,2]
[1,] 1 3
[2,] 2 4
23 Arrays are Created in Column Major
Order > v <- 1:8
> dim(v) <- c(2,2,2)
>v
,,1
[,1] [,2] Start from the last index
[1,] 1 3
[2,] 2 4
,,2 Array elements are accessed
by specifying their index
[,1] [,2] (within square brackets)
[1,] 5 7
[2,] 6 8
> v[2,1,2]
[1] 6
24 The matrix command
A matrix is a 2-D array
There is a fast method of creating a matrix
Use the matrix (data, dim1, dim2) command
Example:
> matrix(1:4, 2, 2)
[,1] [,2]
[1,] 1 3
[2,] 2 4
25 cbind and rbind
mat1 mat2 mat1 mat2
cbind
mat1
mat1 mat2
mat2
rbind
26 Problem: set the diagonal elements of
a matrix to 0
> mat <- matrix(1:16,4,4)
> mat
[,1] [,2] [,3] [,4]
[1,] 1 5 9 13
[2,] 2 6 10 14
[3,] 3 7 11 15
[4,] 4 8 12 16
> indices <- cbind (1:4, 1:4)
> mat[indices] <- 0
> mat
[,1] [,2] [,3] [,4]
[1,] 0 5 9 13
[2,] 2 0 10 14
[3,] 3 7 0 15
[4,] 4 8 12 0
27 Recycling Rule
> cbind (1:4, 1:8)
[,1] [,2]
[1,] 1 1
[2,] 2 2
[3,] 3 3
[4,] 4 4
[5,] 1 5
[6,] 2 6
[7,] 3 7
[8,] 4 8
The smaller structure is replicated to match the
length of the longer structure
Note that the size of the longer structure has to
be a multiple of the size of the smaller structure.
Matrix Operations
28
A * B is a normal element-by-element product
A %*% B is a matrix product
Equation solution:
solve (A, b) (for equations of the form Ax = b)
solve (A) returns the inverse of the matrix
> A <- matrix (1:4, 2, 2)
> b <- 5:6 Solve an equation of the
> solve (A,b) form: Ax = b
[1] -1 2
> solve(A) %*% b
[,1] A-1 * b = x
[1,] -1
[2,] 2
29 Additional Features
nrow (mat) Number of rows in the matrix
ncol (mat) Number of columns in the matrix
Feature Function
Eigen Values eigen
Singular Value Decomposition svd
Least Squares Fitting lsfit
QR decomposition qr
30 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
31 Lists and Data Frames
A list is a heterogeneous data structure
It can contain data belonging to all kinds of types
Example:
> lst <- list (“one”, 1, TRUE)
Elements can be lists, arrays, factors, and normal variables
The components are always numbered
They are accessed as follows: lst[[1]], lst[[2]], lst[[3]]
[[ ... ]] is the operator for accessing an element in a list
32 Named Components
Lists can also have named components
lst <- list(name=“Sofia”, age=29, marks=33.7)
The three components are: lst$name, lst$age, lst$marks
We can also use
lst [[“name”]], lst[[“age”]], lst [[“marks”]]
33 Data Frames columns
rows Data Frame
It is a table in R
> entries <- c(“cars”, “trucks”, “bikes”)
> price <- c (8, 10, 5)
> num <- c (1, 2, 3)
> df <- data.frame(entries, price, num)
> df
entries price num
1 cars 8 1
2 trucks 10 2
3 bikes 5 3
34 Accessing an Element
Can be accessed as a regular array, or as a list
> df[1,2]
Row names, i.e. [1] 8
character values > df[2,]
entries price num
2 trucks 10 2
> df$price
[1] 8 10 5
Summary shows a summary of each variable in the data frame
Feature Function
> summary(df) Show first 6 rows of head(df)
entries price num df
bikes :1 Min. : 5.000 Min. :1.0
cars :1 1st Qu.: 6.500 1st Qu.:1.5 List objects ls()
trucks:1 Median : 8.000 Median :2.0
Remove variables x rm(x,y)
Mean : 7.667 Mean :2.0
3rd Qu.: 9.000 3rd Qu.:2.5
& y from data frame
35 Operations on Data Frames
A data frame can be sorted on the values of a variable,
filtered using values of a variable, and grouped by a variable.
Eg. Filter rows where entries = “cars”
> df[df$entries == "cars",]
entries price num
1 cars 8 1
Group by entries
> aggregate(df,by = list(entries), mean)
Group.1 entries price num
1 bikes NA 5 3
2 cars NA 8 1
3 trucks NA 10 2
36 Reading Data from Files
Reads in a data frame from a file
Steps:
Store the data frame in a file
Read it in
> df <- read.table (“<filename>”)
Access the data frame
37 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
38 Grouping, Loops, Conditional
Execution
R does have support for regular if statements,
while loops, and other conditionals
if statement
if (condition) statement 1 else statement 2. Use {}
for creating grouped statements
The condition should evaluate to a single variable
(not a vector)
Example:
> x <- 3
> if (x > 0) x <- x+ 3 else x <- x + 6
> x
[1] 6
39 For loop
for (var in expr1) {
....
....
}
Example: > for (v in 1:10) print (v)
[1] 1
[1] 2
[1] 3
[1] 4
[1] 5
[1] 6
[1] 7
[1] 8
[1] 9
[1] 10
40 While loop
> while (x[i] < 10) {
+ print (x[i])
+ i <- i + 1
+ }
[1] 1
[1] 2
[1] 3
[1] 4
[1] 5
[1] 6
[1] 7
[1] 8
[1] 9
Use the break statement to exit
a loop
41 Writing one’s own functions
> cube <- function (x) {
+ x * x * x
+ }
> cube(4)
[1] 64
A function takes a list of arguments within ( ... )
To return a value, just print the expression
(without assignment statements)
Function calling convention similar to C
42 Applying a Function
> lapply (1:2,cube)
[[1]]
[1] 1
[[2]]
[1] 8
Apply the cube function to a vector
Applies the function to each and every argument
sapply returns a list
> sapply (1:3, cube)
[1] 1 8 27
43 Named arguments
> fun <- function (x=4, y=3) { x - y }
> fun()
[1] 1
> fun (4,3)
[1] 1
> fun (y=4, x=3)
[1] -1
Possible to specify default values in the function
declaration
If a variable is not specified, the default value is used
We can also specify the values of the variables by the
name of the argument (last line)
44 Scoping in R
> deposit <- function (amt) balance + amt
> withdraw <- function (amt) balance - amt
> balance <- withdraw(10)
> balance <- deposit (20)
> balance
[1] 110
Scope of variables in R
Function arguments (valid only inside the function)
Local variables (valid only inside the function)
Global variables (balance)
45 Functional Programming: Closures
> exponent <- function (n) {
+ power <- function (x) {
+ x ** n
+ }
+ }
> square <- exponent(2)
> square(4)
[1] 16
A function with pre-specified data is called a closure
exponent returns a function power (with n = 2)
source http://adv-r.had.co.nz/Functional-programming.html
46 Example: Numerical Integration
> composite <- function(f, a, b, n = 10,
rule) {
area <- 0
+ points <- seq(a, b, length = n + 1)
+
+ area <- 0
+ for (i in seq_len(n)) { Function for
+ area <- area + rule(f, points[i], numerical
points[i + 1]) integration
+ }
+
+ area
+ }
> midpoint <- function(f, a, b) {
+ (b - a) * f((a + b) / 2) Midpoint rule
function passed + }
as an argument > composite(sin, 0, pi, n = 1000, rule =
𝜋
midpoint)
∫ sin ( 𝑥 ) 𝑑𝑥
[1] 2.00000 0
47 Outline
Variables and Vectors
Factors
Arrays and Matrices
Data Frames
Functions and Conditionals
Graphical Procedures
48 Plotting a Function
A basic 2D plot:
Plot type
vec1 <-cube(seq(1,100,10)) (overplotted)
vec2 <-cube(seq(5,100,10))
plot(vec1, type="o", col="blue“, ylim=c(0,3e5))
title(main=“Plot of Cubes", col.main="red")
To add a line to the same plot:
lines(vec2, type=“o", lty = 2, pch = 22, col=“red“)
Line type: Marker type:
dashed square
To add a legend:
legend(1, max(vec1), c(“vec1",“vec2"), cex=0.8, col=c("blue","red"),
pch=21:22, lty=1:2)
49 Plotting: Linear Regression
library("MASS")
data(cats) # load data
plot(cats$Bwt, cats$Hwt) # scatter plot of cats body weight vs heart rate
M <- lm(formula = cats$Hwt ~ cats$Bwt, data=cats) # fit a linear model
regmodel <- predict(M) # predict values using this model
plot(cats$Bwt, cats$Hwt, pch = 16, cex = 1.3, col = "blue", main = "Heart
rate plotted against body weight of cats", xlab = "Body weight", ylab =
"Heart rate") # scatter plot
abline(M) # plot the regression line
50 Creating 3-D plots
Packages plot3D, ggplot2 contain useful 3D
plotting options
plot3d, scatter3d, surf3d, persp3d are some
of the commonly used plots.
plot3d is from package rgl.
It allows creating interactive 3D plots that can be
rotated using the mouse.
plot3d(x, y, z, col="red", size=3)
51 Creating 3-D plots: surf3D
Surf3d (package: plot3D) allows us to create
surface plots like the one shown below:
#source: http://blog.revolutionanalytics.com/2014/02/3d-
plots-in-r.html
library ('ggplot2')
library(plot3D)
par(mar = c(2, 2, 2, 2))
par(mfrow = c(1, 1))
R <- 3; r <- 2
x <- seq(0, 2*pi,length.out=50)
y <- seq(0, pi,length.out=50)
M <- mesh(x, y)
alpha <- M$x; beta <- M$y
surf3D(x = (R + r*cos(alpha)) * cos(beta),
y = (R + r*cos(alpha)) * sin(beta),
z = r * sin(alpha),
colkey=FALSE,
bty="b2",
main="Half of a Torus")
52 Creating 3-D plots: persp3d
persp3d(package: plot3D) allows us to create
surface plots like the one shown below:
xdim <- 16
newmap <- array(0,dim=c(xdim,xdim))
newmap <- rnorm(256,1,.2)
jet.colors <- colorRampPalette( c("yellow", "red") )
pal <- jet.colors(100)
col.ind <- cut(newmap,100) # colour indices of each point
persp3d(seq(1:xdim),seq(1:xdim),newmap,shade=TRUE,
type="wire", col=pal[col.ind],xlab="",ylab="",zlab="",
cex.axis=1.5,xtics="",aspect=2,zlim=c(0,5))
53
