Introduction to R

• Open Source / Free
• Increasingly popular
• Powerful and Extensible
• Makes reproducible research easy, convenient and diverse visualization options, more statistics than you can shake a stick at, excellent for exploratory data analysis, many support options, often first for cutting-edge techniques, ...
• Available:
  • Download: R / RStudio
  • Data Services lab, fifth floor Bobst
  • Most ITS labs
  • Virtual Computing Lab (VCL) (for students)
  • High Performance Computing (HPC) clusters (requires account)

The Comprehensive R Archive Network (CRAN) hosts this many packages.

This is as of Thu Apr 11 11:30:37 2013.

length(unique(rownames(available.packages())))

## [1] 4332

And there are many more in addition to the ones on CRAN.

• It's not Excel.
• It's not Mathematica/Maple/etc.
• It's not SAS/Stata/SPSS/etc.
• It's not C.
• Defaults to in-memory.
• Often not best for building interactives.

An Integrated Development Environment (IDE) for R. Check it out!

Anything you want to do in R is done by telling R to run a function.

To run a function with no arguments, follow its name with parentheses.

help()

Arguments are passed inside the parentheses. Arguments are usually named, but names can be omitted if it's unambiguous.

help(topic = getwd)
help(getwd)

If you don't include parentheses, R will try to give you the function itself.

help
help.search

Even things that don't look like functions are functions.

5 + 7

## [1] 12

"+"(5,7)

## [1] 12

Arithmetic operations are functions.

Even things that don't look like functions are functions.

":"(1,10)

##  [1]  1  2  3  4  5  6  7  8  9 10

1:10

##  [1]  1  2  3  4  5  6  7  8  9 10

This is a super handy function! It returns a vector.

Convenient short-hand is available for other functions too. Get help fast:

?glm             #  This is identical to: help(glm)

And of course, assign things to variables:

my.object <- 8   #  You will never see the equivalent: "<-"(my.object, 8)



# Okay, comments aren't functions.

42:100

##  [1]  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58
## [18]  59  60  61  62  63  64  65  66  67  68  69  70  71  72  73  74  75
## [35]  76  77  78  79  80  81  82  83  84  85  86  87  88  89  90  91  92
## [52]  93  94  95  96  97  98  99 100

The numbers in brackets tell you the position in the vector at the start of the line. So:

42

## [1] 42

2, 4      #  this will fail

c(2, 4)   #  this will make a vector containing first 2 then 4

Very often you will want to pass one vector as an argument to a function.

mean(2, 4)      #  this passes the function two arguments,
                #   a vector containing 2 and a vector containing 4

mean(c(2, 4))   #  this passes the function one argument,
                #   a vector containing first 2 then 4

This kind of thing is common in R and an easy way to make a mistake.

class(TRUE); class(T); class(FALSE); class(F);              #  logical
class(1:10); class(42L);                                    #  integer
class(42); class(3.7); class(5e7); class(1/89)              #  numeric
class("Aaron"); class("cow"); class("123"); class("TRUE")   #  character

# And then there are these guys...
class(factor(c("red", "green", "blue")))                    #  factor
class(factor(c("medium", "small", "small", "large"),
             levels=c("small", "medium", "large"),
             ordered=TRUE))                                 #  ordered factor

Vectors have exactly one class, and are joined by the c() function.

c(9, 7, TRUE, FALSE)
c(9, 7, TRUE, FALSE, "cow")

Other things: NA (missing), NULL (not a thing), NaN (sqrt(-1)), Inf (1/0).

Most operations happen element-wise.

c(1, 2, 3, 4) + c(100, 1000, 10000, 10000)

## [1]   101  1002 10003 10004

If the vectors have different lengths, they shorter one gets 'recycled'.

c(1, 2, 3, 4) + c(100, 1000)

## [1]  101 1002  103 1004

What will happen with these?

c(1, 2) * c(4, 5, 6)

1 + 1:10

1:10 / 10

1:10 < 5

c(1, 2) * c(4, 5, 6)

## Warning: longer object length is not a multiple of shorter object length

## [1]  4 10  6

1 + 1:10

##  [1]  2  3  4  5  6  7  8  9 10 11

1:10 / 10

##  [1] 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

1:10 < 5

##  [1]  TRUE  TRUE  TRUE  TRUE FALSE FALSE FALSE FALSE FALSE FALSE

my.vector <- 101:105
my.vector

## [1] 101 102 103 104 105

names(my.vector) <- c("a", "b", "c", "d", "e")  # don't be scared!
my.vector

##   a   b   c   d   e 
## 101 102 103 104 105

my.vector[c(2, 4)]                             # by index numbers

##   b   d 
## 102 104

my.vector[c('c', 'e')]                         # by names

##   c   e 
## 103 105

my.vector[c(TRUE, FALSE, TRUE, FALSE, TRUE)]   # with logicals

##   a   c   e 
## 101 103 105

(my.numbers <- sample(1:10, 20, replace = TRUE))

##  [1] 10 10  3  9  7  6  8  2  7  8  5  8 10  3  5 10 10  2  5  6

How can we get just the entries less than five?

my.numbers < 5

##  [1] FALSE FALSE  TRUE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE
## [12] FALSE FALSE  TRUE FALSE FALSE FALSE  TRUE FALSE FALSE

my.numbers[my.numbers < 5]

## [1] 3 2 3 2

length(my.vector)   #  How long is my vector?

## [1] 5

sum(my.vector)      #  What if I add up the numbers in my vector?

## [1] 515

sum(my.vector < 4)  #  Alternative: length(my.vector[my.vector < 4])

## [1] 0

• Matrices are vectors with a number of columns and a number of rows, which should all jive.
  • Multiplication is element-wise for *, matrix-wise for %*%.
• Lists are like vectors where each element could be itself a vector.
  • Compare c(1:3, 4) with list(1:3, 4).
• Data frames are lists with every vector equal length, and you get row names and column names.

(my.data <- read.csv("http://bit.ly/NYUdataset"))

##    id gender age time health1 health2 health3 health4 health5 health6
## 1   1      M  51   15       1       4       2       1       4       5
## 2   2      F  35   30       2       3       3       2       3       4
## 3   3      F  29   25       5       2       4       2       1       3
## 4   4      M  21   40       5       1       5       4       2       1
## 5   5      M  56   30       2       4       2       4       3       3
## 6   6      M  72   10       1       5       4       2       4       5
## 7   7      F  46   20       2       5       3       1       3       4
## 8   8      M  33   25       5       2       4       5       2       1
## 9   9      F  36   30       3       3       4       5       2       2
## 10 10      M  42   20       3       3       3       4       2       4
## 11 11      F  41   10       2       4       3       3       3       3
## 12 12      F  57   45       1       4       2       1       5       5
## 13 13      M  30   10       3       2       3       4       1       3
## 14 14      F  48   15       5       3       3       4       2       2
## 15 15      M  32    0       4       2       4       3       2       2

str(my.data)
summary(my.data)

You can access a particular vector in a list or data frame in several ways:

my.data$gender
my.data[[2]]
my.data[["gender"]]
with(my.data, gender)

You can subset using [row(s), column(s)], both parts just like selecting from a single vector.

my.data[2, "age"]

## [1] 35

How can we select the times for females?

How can we select the times for females?

my.data[my.data$gender == "F", "time"]

Other options:

my.data$time[my.data$gender == "F"]

subset(my.data, gender == "F", select = "time")

To add / compute / make a new column, just assign to it:

my.data$number.five <- 5
my.data$mean.1.2 <- my.data$health1 + my.data$health2
my.data$health <- rowMeans(my.data[5:10])

To drop / delete / remove a column, you have options:

my.data$number.five <- NULL         #  remove from the data frame 'in place'
my.new.data <- my.data[1:10]        #  make a new smaller data frame
my.new.data <- my.data[-c(11,12)]   #  same as last

mean(my.data$age)
sd(my.data$age)
cor(my.data[5:10])
table(my.data$gender)
table(my.data$health3, my.data$gender)
chisq.test(my.data$health3, my.data$gender)
with(my.data, t.test(health1, health2))
my.model <- lm(health1 ~ age + gender, data = my.data)
summary(my.model)
confint(my.model)
aov(my.model)
aov(health1 ~ age + gender, data = my.data)

with(my.data, barplot(table(gender)))
plot(my.data$age)
hist(my.data$age)
hist(my.data$age, col = "cornflowerblue", breaks = 20, xlab = "Age", main = "Participants")
boxplot(my.data$age)
with(my.data, boxplot(age ~ gender))
with(my.data, plot(health1, health2))
with(my.data, plot(health1, health2, pch = 19))
with(my.data, plot(jitter(health1), jitter(health2)))
with(my.data, plot(jitter(health1), jitter(health2), pch = 20, col = rainbow(15), 
    xlab = "Monkeys eaten", ylab = "Number of cheeses", main = "Absolute Power (Ninjas)"))
pairs(my.data[5:10])
plot(my.model)

There are many packages available on the Comprehensive R Archive Network (CRAN) which can be easily installed and loaded into R. One very popular package is ggplot2, a graphing library.

install.packages('ggplot2')  # Do this once per machine.
library(ggplot2)             # Do this once per R session.

After installing and loading a package, you can use the functions it provides.

qplot(x = carat, y = price, color = cut, data = diamonds) + theme_bw()

• Try R: A free online interactive tutorial
• A Beginners Guide to R (book)
• The Art of R Programming (book)