basic r - ucla statistical consulting...

UCLA Department of StatisticsStatistical Consulting Center

Basic R

Mark [email protected]

October 1, 2010

Mark Nakamura [email protected]

Basic R UCLA SCC

Acknowledgements

This mini-course was developed by Irina Kukuyeva andmodified by Brigid Wilson for this presentation. Many thanksto them for sharing her work.


Basic R UCLA SCC

Outline

I. Preliminaries

II. Variable Assignment

III. Working with Vectors

IV. Working with Matrices

V. From Vectors to Matrices

VI. More on Handling Missing Data

VII. The Help System

VIII. Datasets in R

IX. Overview of Plots

X. R Environment

XI. Common Bugs and Fixes

XII. Online Resources for R

XIII. Exercises

XIV. Upcoming Mini-Courses


Basic R UCLA SCC

Part I

Preliminaries


Basic R UCLA SCC

Software Installation

Installing R on Mac

1 Go tohttp://cran.r-project.organd select MacOS X.

2 Select to download thelatest version: 2.10.1

3 Install and Open. The Rwindow should look likethis:


Basic R UCLA SCC

Software Installation

Installing R on Windows

1 Go tohttp://cran.r-project.organd select Windows.

2 Select base to install theR system.

3 Click on the largedownload link. There isother informationavailable on the page.

4 Install and Open. The Rwindow should look likethis:


Basic R UCLA SCC

Part II

Variable Assignment


Basic R UCLA SCC

Creating Variables

Creating Variables I

To use R as a calculator, type an equation and hitENTER. (Note how R prints the result.) Your outputshould look like this:

1 2 + 52 [1] 7


Basic R UCLA SCC

Creating Variables

Creating Variables II

To create variables in R, use either <- or =:

1 # Approach 1

2 a=53 a4 [1] 55 # Approach 2

6 b<-57 b8 [1] 5


Basic R UCLA SCC

Creating Variables

Creating Variables IIICaution!

Be careful when using <- to compare a variable with anegative number!

1 #Assign a value to a

2 a<--23 #Is a less than -5?

4 a<-55 a6 [1] 5 #Expected FALSE


Basic R UCLA SCC

Creating Variables

Creating Variables IV

Use spaces so that R will not be confused. It is better to useparentheses instead.

1 a <- 52 a < -23 [1] FALSE


Basic R UCLA SCC

Creating Variables

Creating Variables VCaution!

It is important not to name your variables after existingvariables or functions. For example, a bad habit is to nameyour data frames data. data is a function used to load somedatasets.If you give a variable the same name as an existingconstant, that constant is overwritten with the value ofthe variable. So, it is possible to define a new value for !.


Basic R UCLA SCC

Creating Variables

Creating Variables VICaution!

On the other hand, if you give a variable the same name as anexisting function, R will treat the identifier as a variable if usedas a variable, and will treat it as a function when it is used asa function:

c <- 2 #typing c yields "2"c(c,c) #yields a vector containing two 2s.


Basic R UCLA SCC

Creating Variables

Creating Variables VII

Caution!

As we have seen, you can get away with using the same namefor a variable as with an existing function, but you will be introuble if you give a name to a function and a function withthat name already exists.


Basic R UCLA SCC

Part III

Working with Vectors


Basic R UCLA SCC

Creating Vectors

Creating Vectors I

Scalars are the most basic vectors. To create vectors of lengthgreater than one, use the concatenation function c():

1 d=c(3,4,7); d2 [1] 3 4 7

The More You Know...

The semicolon ; is used to combine multiple statements onone line.


Basic R UCLA SCC

Creating Vectors

Creating Vectors II

To create a null vector:

1 x=c(); x2 NULL


Basic R UCLA SCC

Creating Vectors

Creating Vectors III

Creating a vector with equal spacing, use the sequencefunction seq():

1 e=seq(from=1, to=3, by=0.5); e2 [1] 1.0 1.5 2.0 2.5 3.0

Creating a vector of a given length, use the repeat functionrep():

1 f=rep(NA, 6); f2 [1] NA NA NA NA NA NA


Basic R UCLA SCC

Some Vector Functions

Some Useful Vector Functions I

To find the length of the vector, use length():

1 length(d)2 [1] 3

To find the maximum value of the vector, use the maximumfunction max():

1 max(d)2 [1] 7


Basic R UCLA SCC


Some Useful Vector Functions II

To find the minimum value of the vector, use the minimumfunction min():

1 min(d)2 [1] 3

To find the mean of the vector, use mean():

1 mean(d)2 [1] 4.666667


Basic R UCLA SCC


Some Useful Vector Functions III

To sort the vector, use sort():

1 g<-c(2,6,7,4,5,2,9,3,6,4,3)2 sort(g, decreasing=TRUE)3 [1] 9 7 6 6 5 4 4 3 3 2 2

Caution!

Although T and F work in place of TRUE and FALSE, it is notrecommended.


Basic R UCLA SCC


Some Useful Vector Functions IV

To find the unique elements of the vector, use unique():

1 unique(g)2 [1] 2 6 7 4 5 9 3

Alternatively, to find the elements of the vector that repeat,use duplicated():

1 duplicated(g)2 [1] FALSE FALSE FALSE FALSE FALSE TRUE3 [7] FALSE FALSE TRUE TRUE TRUE


Basic R UCLA SCC


Some Useful Vector Functions V

To determine if a value is missing (NA), use is.na. This isuseful for finding missing values and removing them, or doingsomething else with them.

1 a <- c(1,2,3,NA ,6)2 is.na(a)3 [1] FALSE FALSE FALSE TRUE FALSE

But some functions do not tolerate missing values.


Basic R UCLA SCC


Some Useful Vector Functions VICaution!

mean(a)[1] NAmean(a, na.rm=TRUE)[1] 1.5


Basic R UCLA SCC


Some Useful Vector Functions VII

To get the number of missing values in a vector,

1 sum(is.na(a))2 [1] 1

There are other ways to handle missing values. See?na.action.


Basic R UCLA SCC


Some Useful Vector Functions VIII

One final common function you can use on vectors (and otherobjects) is summary.

1 summary(a)

Min. 1st Qu. Median Mean 3rd Qu. Max.1.00 1.75 2.50 3.00 3.75 6.00NA’s1.00

There are many, many other functions you can use on vectors!


Basic R UCLA SCC

Comparisons in R

Comparisons in R

Symbol Meaning! logical NOT& logical AND— logical OR< less than<= less than or equal to> greater than>= greater than or equal to== logical equals! = not equal


Basic R UCLA SCC

Subsetting with Vectors

Subsetting with Vectors I

To find out what is stored in a given element of the vector,use [ ]:

1 d[2]2 [1] 4

To see if the elements of a vector equal a certain number,use ==:

1 d==32 [1] TRUE FALSE FALSE


Basic R UCLA SCC


Subsetting with Vectors II

To see if any of the elements of a vector do not equal acertain number, use !=:

1 d!=32 [1] FALSE TRUE TRUE


Basic R UCLA SCC


Subsetting with Vectors III

To obtain the element number of the vector when a conditionis satisfied, use which():

1 which(d==4)2 [1] 2

To store the result, type: a=which(d==4); a


Basic R UCLA SCC


Subsetting with Vectors IV

We can also tell R what we do not want when subsetting byusing the minus - sign. To obtain everything but the 2ndelement,

1 d <- seq(1,10,2)2 d[-2]3 [1] 1 5 7 9


Basic R UCLA SCC


Subsetting with Vectors V

We can use subsetting to explicitly tell R what observations wewant to use. To get all elements of d greater than or equal to2,

1 d[d >= 2]2 [1] 3 5 7 9

R will return values of d where the expression within bracketsis TRUE. Think of these statements as: “give me all d suchthat d ! 2.”


Basic R UCLA SCC


Exercise 1

Create a vector of the positive odd integers less than 100

Remove the values greater than 60 and less than 80

Find the variance of the remaining set of values


Basic R UCLA SCC

Part IV

Working with Matrices


Basic R UCLA SCC

Creating Matrices

Creating Matrices I

To create a matrix, use the matrix() function:

1 mat <-matrix (10:15 , nrow=3, ncol =2); mat2 [,1] [,2]3 [1,] 10 134 [2,] 11 145 [3,] 12 15


Basic R UCLA SCC

Some Matrix Functions

Some Useful Matrix Functions I

To add two matrices, use +

1 mat+mat2 [,1] [,2]3 [1,] 20 264 [2,] 22 285 [3,] 24 30


Basic R UCLA SCC


Some Useful Matrix Functions II

To find the transpose of a matrix, use t():

1 t(mat)2 [,1] [,2] [,3]3 [1,] 10 11 124 [2,] 13 14 15


Basic R UCLA SCC


Some Useful Matrix Functions III

To find the dimensions of a matrix, use dim():

1 dim(mat)2 [1] 3 2

Alternatively, we can find the rows and columns of the matrix,by nrow() and ncol().


Basic R UCLA SCC


Some Useful Matrix Functions IV

To multiply two matrices, use %*%.Note: If you use * instead, you will be performing matrixmultiplication element-wise.

1 mat%*%t(mat)2 [,1] [,2] [,3]3 [1,] 269 292 3154 [2,] 292 317 3425 [3,] 315 342 369


Basic R UCLA SCC

Subsetting with Matrices

Subsetting with Matrices I

To see what is stored in the first element of the matrix,use [ ]:

1 mat[1,1]2 [1] 10

To see what is stored in the first row of the matrix:

1 mat[1,]2 [1] 10 13


Basic R UCLA SCC


Subsetting with Matrices II

To see what is stored in the second column of the matrix:

1 mat[, 2]2 [1] 13 14 15


Basic R UCLA SCC


Subsetting with Matrices III

To extract elements 1 and 3 from the second column, use c()and [ ]:

1 mat[c(1,3), 2]2 [1] 13 15


Basic R UCLA SCC

Part V

From Vectors to Matrices


Basic R UCLA SCC

Creating Matrices from Vectors

Creating Matrices from Vectors I

To stack two vectors, one below the other, use rbind():

1 mat1 <-rbind(d,d); mat12 [,1] [,2] [,3]3 d 3 4 74 d 3 4 7


Basic R UCLA SCC

Creating Matrices from Vectors

Creating Matrices from Vectors II

To stack two vectors, one next to the other, use cbind():

1 mat2 <-cbind(d,d); mat22 d d3 [1,] 3 34 [2,] 4 45 [3,] 7 7


Basic R UCLA SCC

Part VI

More on Handling Missing Data


Basic R UCLA SCC

Missing Data in Matrices

Missing Data in Matrices I

Start by creating a matrix with missing data:

1 h=matrix(c(NA ,3,1,7,-8,NA), nrow=3,ncol=2, byrow=TRUE); h

2 [,1] [,2]3 [1,] NA 34 [2,] 1 75 [3,] -8 NA


Basic R UCLA SCC


Missing Data in Matrices II

To see if any of the elements of a vector are missing useis.na():

1 is.na(h)2 [,1] [,2]3 [1,] TRUE FALSE4 [2,] FALSE FALSE5 [3,] FALSE TRUE


Basic R UCLA SCC


Missing Data in Matrices III

To see how many missing values there are, use sum() andis.na() (TRUE=1, FALSE=0):

1 sum(is.na(h))2 [1] 2

To obtain the element number of the matrix of the missingvalue(s), use which() and is.na():

1 which(is.na(h))2 [1] 1 6


Basic R UCLA SCC


Missing Data in Matrices IV

To keep only the rows without missing value(s), usena.omit()

1 na.omit(h)2 [,1] [,2]3 [1,] 1 74 attr(,"na.action")5 [1] 1 36 attr(,"class")7 [1] "omit"


Basic R UCLA SCC


Exercise 2

Find the matrix product of A and B if

Matrix A=!

"2 3 71 6 23 5 1

#

$

Matrix B=!

"3 2 90 7 85 8 2

#

$


Basic R UCLA SCC

Part VII

The Help System


Basic R UCLA SCC

Help with a Function

Help with a Function I

To get help with a function in R, use ? followed by the nameof the function.

1 ?read.table

help(function name) also works.


Basic R UCLA SCC

Help with a Function

Help with a Function II


Basic R UCLA SCC

Help with a Package

Help with a Package I

To get help with a package, use help(package="name").

1 help(package="MASS")


Basic R UCLA SCC

Help with a Package

Help with a Package II


Basic R UCLA SCC

Searching for Help

Searching for Help I

To search R packages for help with a topic, usehelp.search().

1 help.search("regression")


Basic R UCLA SCC

Searching for Help

Searching for Help II


Basic R UCLA SCC

Part VIII

Datasets in R


Basic R UCLA SCC

Importing Datasets into R

Data from the Internet I

When downloading data from the internet, useread.table(). In the arguments of the function:

header if TRUE, tells R to include variables names whenimporting

sep tells R how the entires in the data set are separatedsep="," when entries are separated by COMMASsep="\t" when entries are separated by TABsep=" " when entries are separated by SPACE


Basic R UCLA SCC


Data from the Internet II

1 stock.data <-read.table("http://www.google.com/finance/historical?q=NASDAQ:AAPL&output=csv", header=TRUE , sep=",")


Basic R UCLA SCC


Importing Data from Your Computer I

1 Check what folder R is working with now:

1 getwd()

2 Tell R in what folder the data set is stored (if di!erentfrom (1)). Suppose your data set is on your desktop:

1 setwd("~/Desktop")

3 Now use the read.table() command to read in thedata, substituting the name of the file for the website.


Basic R UCLA SCC


Using Data Available in R I

To use a data set available in one of the R packages, installthat package (if needed). Load the package into R, using thelibrary() function.

1 library(alr3)

Extract the data set you want from that package, using thedata() function. In our case, the data set is called UN2.

1 data(UN2)


Basic R UCLA SCC

Working with Datasets in R

Working with Datasets in R I

To use the variable names when working with data, useattach():

1 data(UN2)2 attach(UN2)


Basic R UCLA SCC


Working with Datasets in R II

After the variable names have been ”attached”, to see thevariable names, use names():

1 names(UN2)

To see the descriptions of the variables, use ?:

1 ?UN2


Basic R UCLA SCC


Working with Datasets in R III

After modifying variables, use detach() and attach() tosave the results:

1 # Make a copy of the data set

2 UN2.copy <-UN23 detach(UN2)4 attach(UN2.copy)5 # Change the 10th observation for

logFertility

6 UN2.copy[10, 2]<-999


Basic R UCLA SCC


Working with Datasets in R IV

To get an overview of the data sets and its variables, use thesummary() function:

1 # Check that the change has been made

2 summary(UN2)3 summary(UN2.copy)


Basic R UCLA SCC


Working with Datasets in R VCaution!

Avoid using attach() if possible. Many strange things canoccur if you accidentally attach the same data frame multipletimes, or forget to detach. Instead, you can refer to a variableusing $. To access the Locality variable in data frame UN2,use UN2$Locality. You can also get around this by using thewith function or if your function of choice takes dataargument.

“attach at your own risk!”


Basic R UCLA SCC


Working with Datasets in R VI

To get the mean of all the variables in the data set, usemean():

1 mean(UN2 ,na.rm=TRUE)2 logPPgdp logFertility Purban3 10.993094 1.018016 55.538860


Basic R UCLA SCC


Working with Datasets in R VII

To get the correlation matrix of all the (numerical) variables inthe data set, use cor():

1 cor(UN2 [ ,1:2])2 logPPgdp logFertility3 logPPgdp 1.000000 -0.6776044 logFertility -0.677604 1.000000

Can similarly obtain the variance-covariance matrix usingvar().


Basic R UCLA SCC


Exercise 3

Load the Animals dataset from the MASS package

Examine the documentation for this dataset

Find the correlation coe"cient of brain weight and bodyweight in this dataset


Basic R UCLA SCC

Part IX

Overview of Plots in R


Basic R UCLA SCC

Creating Plots

Basic scatterplot I

To make a plot in R, you can use plot():

1 plot(x=UN2$logPPgdp ,2 y = UN2$logFertility ,3 main = "Fertility vs. PerCapita GDP",4 xlab = "log PerCapita GDP , in $US",5 ylab = "Log Fertility")


Basic R UCLA SCC

Creating Plots

Basic scatterplot II

8 10 12 14

0.0

0.5

1.0

1.5

2.0

Fertility vs. PerCapita GDP

log PerCapita GDP, in $US

Log F

ert

ility


Basic R UCLA SCC

Creating Plots

Histogram I

To make a histogram in R, you can use hist():

1 hist(UN2$logPPgdp ,2 main = "Distribution of PerCapita GDP",3 xlab = "log PerCapita GDP")


Basic R UCLA SCC

Creating Plots

Histogram II

Distribution of PerCapita GDP

log PerCapita GDP

Frequency

6 8 10 12 14 16

05

10

15

20

25

30


Basic R UCLA SCC

Creating Plots

Boxplot I

To make a boxplot in R, you can use boxplot():

1 boxplot(UN2$logPPgdp ,2 main = "Boxplot of PerCapita GDP")


Basic R UCLA SCC

Creating Plots

Boxplot II

810

12

14

Boxplot of PerCapita GDP


Basic R UCLA SCC

Creating Plots

Matrix of Scatterplots I

To make scatterplots of all the numeric variables in yourdataset in R, you can use pairs():

1 pairs(UN2)


Basic R UCLA SCC

Creating Plots

Matrix of Scatterplots II

logPPgdp

0.0 0.5 1.0 1.5 2.0

810

12

14

0.0

0.5

1.0

1.5

2.0

logFertility

8 10 12 14 20 40 60 80 100

20

40

60

80

100

Purban


Basic R UCLA SCC

Creating Plots

Overlaying with points I

To add more points to an existing plot, use points().Here,we will first plot fertility vs. PerCapita GDP first where %urban is less than 50.

1 attach(UN2)2 plot(logPPgdp[Purban < 50],3 logFertility[Purban < 50],4 main = "Fertility vs. PPGDP , by % Urban",5 xlab = "log PPGDP",6 ylab = "log Fertility")


Basic R UCLA SCC

Creating Plots

Overlaying with points II

We then add in the points where % urban is more than 50 andmark these points with a di!erent color.

1 points(logPPgdp[Purban >= 50],2 logFertility[Purban >= 50],3 col = "red")


Basic R UCLA SCC

Creating Plots

Overlaying with points III

8 10 12 14

0.5

1.0

1.5

2.0

Fertility vs. PPGDP, by % Urban

log PPGDP

log F

ert

ility


Basic R UCLA SCC

Creating Plots

Overlaying

Caution!

Once a plot is constructed using plot, whatever is containedin the plot cannot be modified. To overlay things on arendered plot, use one of the following

1 abline - add a line with slope b, intercept a orhorizontal/vertical.

2 points - add points.3 lines - add lines.


Basic R UCLA SCC

Saving Plots as a PDF

Saving Plots as a PDF I

Note: The files will be saved in the folder specified withsetwd().

To save a plot in R as a PDF, you can use pdf():

1 pdf("myplot.pdf")2 pairs(UN2)3 dev.off()


Basic R UCLA SCC

Part X

R Environment


Basic R UCLA SCC

Exploring R Objects

Exploring R Objects I

To see the names of the objects available to be saved (in yourcurrent workspace), use ls().

1 ls()

[1] "UN2" "a" "b" "d" "data" "e" "f" "h" "mat1" "mat2"


Basic R UCLA SCC

Exploring R Objects

Exploring R Objects II

To remove objects from your workspace, use rm().

1 rm(d)2 ls()

[1] "UN2" "a" "b" "data" "e" "f" "h" "mat1" "mat2"


Basic R UCLA SCC

Exploring R Objects

Exploring R Objects III

To remove all the objects from your workspace, type:

1 rm(list=ls())2 ls()

character(0)


Basic R UCLA SCC

Saving and Loading R Objects

Saving and Loading R Objects I

To save (to the current directory) all the objects in theworkspace, use save.image().

1 save.image("basicR.RData")

To load (from the current directory), use load().

1 load("basicR.RData")


Basic R UCLA SCC

Saving and Loading R Objects

Saving and Loading R Objects I

To save (to the current directory) a single object in theworkspace, use save().

1 save(stock.data ,file="stocks.RData")

To load (from the current directory), use load().

1 load("stocks.RData")


Basic R UCLA SCC

Exporting R Objects to Other Formats I

To save (to the current directory) certain objects in theworkspace to be used in Excel, use write.csv().

1 write.csv(stock.data ,2 file="stockdata.csv")


Basic R UCLA SCC

Saving R Commands

Saving R Commands I

To see all of the commands you typed in an R session,click on the Yellow and Green Tablet


Basic R UCLA SCC

Saving R Commands

Saving R Commands IITo save all of the commands you typed in an R session,use:

1 savehistory(file="history.log")


Basic R UCLA SCC

Saving R Commands

Saving R Commands III

Alternatively, use a .r file to store your commands.1 Go to: File -> New Document2 Type your commands3 Save the file as "code.r"4 Go back to the R Console5 To run all the commands, use:

1 source("code.r")

The More You Know...

Use the # sign to write comments in your code. Use them!


Basic R UCLA SCC

Part XI

Common Bugs and Fixes


Basic R UCLA SCC

Syntax Error

Error: syntax error

Possible causes:

Incorrect spelling (of the function, variable, etc.)

Including a ”+” when copying code from the Console

Having an extra parenthesis at the end of a function

Having an extra bracket when subsetting


Basic R UCLA SCC

Trailing +

Trailing +

Possible causes:

Not closing a function call with a parenthesis

Not closing brackets when subsetting

Not closing a function you wrote with a squiggly brace

You can escape this sticky situation by hitting the ESCAPEkey to exit your command.


Basic R UCLA SCC

Error When Performing Operations

Error in ... : requires numeric

matrix/vector arguments

Possible causes:1 Objects are data frames, not matrices2 Elements of the vectors are characters

Possible solutions:1 Coerce (a copy of) the data set to be a matrix, with the

as.matrix() command2 Coerce (a copy of) the vector to have numeric entries,

with the as.numeric() command


Basic R UCLA SCC

Part XII

Online Resources for R


Basic R UCLA SCC

CRAN

http://cran.stat.ucla.edu/


Basic R UCLA SCC

R-Seek Search Engine

http://www.rseek.org


Basic R UCLA SCC

UCLA Statistics Bootcamp Resources

R Bootcamp is a day-long introduction to R. Handouts anddatasets from Bootcamp 2008 can be found on Ryan Rosario’swebsite: http://www.stat.ucla.edu/"rosario/boot08/


Basic R UCLA SCC

UCLA Statistics Information Portal

http://info.stat.ucla.edu/grad/


Basic R UCLA SCC

UCLA Statistical Consulting Center E-consultingand Walk-in Consulting

http://scc.stat.ucla.edu


Basic R UCLA SCC

Part XIII

Exercises


Basic R UCLA SCC

Exercise 1

Create a vector of the positive odd integers less than 100

Remove the values greater than 60 and less than 80

Find the variance of the remaining set of values


Basic R UCLA SCC

Exercise 2

Find the matrix product of A and B if

Matrix A=!

"2 3 71 6 23 5 1

#

$

Matrix B=!

"3 2 90 7 85 8 2

#

$


Basic R UCLA SCC

Exercise 3

Load the Animals dataset from the MASS package

Examine the documentation for this dataset

Find the correlation coe"cient of brain weight and bodyweight in this dataset


Basic R UCLA SCC

Solutions I

1 e1 <- seq(from = 1, to = 100, by = 2)2 e1.2 <- e1[e1 <= 60 | e1 >= 80]3 var(e1.2)4 [1] 931.282


Basic R UCLA SCC

Solutions II

1 A <- matrix(c(2, 3, 7, 1, 6, 2, 3, 5, 1),nrow = 3, byrow = TRUE)

2 B <- matrix(c(3, 2, 9, 0, 7, 8, 5, 8, 2),nrow = 3, byrow = TRUE)

3 A%*%B4 [,1] [,2] [,3]5 [1,] 41 81 566 [2,] 13 60 617 [3,] 14 49 69


Basic R UCLA SCC

Solutions III

1 library(MASS)2 data(Animals)3 ?Animals4 cor(Animals)5 body brain6 body 1.000000000 -0.0053411637 brain -0.005341163 1.000000000


Basic R UCLA SCC

Part XIV

Upcoming Mini-Course


Basic R UCLA SCC

Upcoming Mini-Courses

April 7 - R Programming II: Data Manipulation and Functions

April 12 - LaTex I: Writing a Document, Paper, or Thesis

April 14 - LaTeX II: Bibliographies, Style and Math in LaTeX

April 19 - LaTeX III: Sweave, Embedding R in LaTeX

For a schedule of all mini-courses o!ered please visithttp://scc.stat.ucla.edu/mini-courses.


Basic R UCLA SCC

basic r - ucla statistical consulting...

Documents