Introduction to R

• A statistical analysis environment
• A calculator!

1 + 2

[1] 3

4 * (1 - 2)

[1] -4

2^10

[1] 1024

A storage for a value or calculation

x <- 1
x

[1] 1

x <- x + 1
x

[1] 2

y <- 3 
z <- x + y
z

[1] 5

A variable can also store a vector of values.

x <- c(4, 4, 1, 5, 100)
x

[1]   4   4   1   5 100

y <- c("abc", "hello")
y

[1] "abc"   "hello"

• R functions are just like mathematical functions \( y = f(x) \) where they take some input \( x \) and return some calculated value.
• We'll use quite a few that are provided with R
• We can write our own

f <- function(x, y) { x^2 + y }
f(2, 3)

[1] 7

x <- f(1, 10)
x

[1] 11

• In some cases we will need to use a “package” which is a collection of functions and data packaged together.
  
• The functions in these packages will not be available for us to use by default, so we need to load them.
• For example, to load the foreign package we use the command

library(foreign)

It is a useful function to get a quick description of data

x <- c(1, 2, 100, 33, -34)
summary(x)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  -34.0     1.0     2.0    20.4    33.0   100.0 

• To simplify our work with R we will use a web-accessible version of RStudio (you do not need to install anything)
• Refer to the relevant post on Piazza for login information
• Each user will have their own home directory on the server where they will save all their work (/home/<username>/)

• We will use projects to help us organize work on assignments
• For each new assignment select New Project from the File Menu
• Create the new project in a new directory and give it a meaningful name such as “assignment1”
• The actual R code we will write will be save in script files, stored in their project directories

• In many cases we will not find it convenient to work on the commandline all the time; we might have many commands we want to execute
• A script is a simple text file in which you list the commands you want executed, one after another
• Once you are done with an assignment's script and want to turn it in, click on the Compile HTML Notebook from R Script button on the upper right side of the script's pane. You can then print the output HTML file that will be displayed.

• All data used in this class will be stored at /usr/data/ec414/

For example, suppose want to load the CPS data used in chapter 3 of the text book

load("/usr/data/ec414/cps_ch3.RData")

If we execute the ls functions to display the variables in the workspace we will see

ls()

[1] "cps.ch3" "f"       "x"       "y"       "z"      

There are several functions we can use to learn more about the data we just loaded

str(cps.ch3)

'data.frame':   15393 obs. of  3 variables:
 $ a.sex: int  1 1 1 2 1 2 2 1 1 1 ...
 $ year : int  1992 1992 1992 1992 1992 1992 1992 1992 1992 1992 ...
 $ ahe08: num  17.2 15.3 22.9 13.3 22.1 ...

summary(cps.ch3)

     a.sex           year          ahe08     
 Min.   :1.00   Min.   :1992   Min.   : 2.0  
 1st Qu.:1.00   1st Qu.:1996   1st Qu.:15.3  
 Median :1.00   Median :2000   Median :20.5  
 Mean   :1.48   Mean   :2001   Mean   :22.4  
 3rd Qu.:2.00   3rd Qu.:2004   3rd Qu.:27.4  
 Max.   :2.00   Max.   :2008   Max.   :82.4  

• Notice that a couple of columns (or variables) cps.ch3 are stored as integers (a.sex and year)
• We would like to have them stored in the dataset (or data.frame) as categorical variables
• We do that using the function factor

cps.ch3$a.sex <- factor(cps.ch3$a.sex, levels=c(1,2), labels=c("male", "female"))
cps.ch3$year <- factor(cps.ch3$year)
str(cps.ch3)

'data.frame':   15393 obs. of  3 variables:
 $ a.sex: Factor w/ 2 levels "male","female": 1 1 1 2 1 2 2 1 1 1 ...
 $ year : Factor w/ 5 levels "1992","1996",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ ahe08: num  17.2 15.3 22.9 13.3 22.1 ...

Using the data we just loaded, we want to calculate the sample mean and variance of average hourly earnings (ahe08)

mean(cps.ch3$ahe08)

[1] 22.4

var(cps.ch3$ahe08)

[1] 108.4

sd(cps.ch3$ahe08) # standard deviation

[1] 10.41

But since we want to compare average wages between men and women, we want to calculate sample statistics for each of these groups

male.cps <- subset(cps.ch3, a.sex == "male")
str(male.cps)

'data.frame':   8008 obs. of  3 variables:
 $ a.sex: Factor w/ 2 levels "male","female": 1 1 1 1 1 1 1 1 1 1 ...
 $ year : Factor w/ 5 levels "1992","1996",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ ahe08: num  17.2 15.3 22.9 22.1 36.1 ...

female.cps <- subset(cps.ch3, a.sex == "female")
str(female.cps)

'data.frame':   7385 obs. of  3 variables:
 $ a.sex: Factor w/ 2 levels "male","female": 2 2 2 2 2 2 2 2 2 2 ...
 $ year : Factor w/ 5 levels "1992","1996",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ ahe08: num  13.28 12.17 21.07 7.82 18.61 ...

To get an more specific subset, for example, female and surveyed in 1992

female.92.cps <- subset(cps.ch3, a.sex == "female" & year == "1992")
str(female.92.cps)

'data.frame':   1368 obs. of  3 variables:
 $ a.sex: Factor w/ 2 levels "male","female": 2 2 2 2 2 2 2 2 2 2 ...
 $ year : Factor w/ 5 levels "1992","1996",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ ahe08: num  13.28 12.17 21.07 7.82 18.61 ...

summary(female.92.cps)

    a.sex        year          ahe08      
 male  :   0   1992:1368   Min.   : 2.61  
 female:1368   1996:   0   1st Qu.:14.75  
               2000:   0   Median :18.65  
               2004:   0   Mean   :20.05  
               2008:   0   3rd Qu.:24.40  
                           Max.   :66.37  

• To plot data in the course we are going to mainly rely on the ggplot2 package.

library(ggplot2)

• The main function we are going to use from this package is qplot

A boxplot is useful at looking at the different distribution of average hourly earnings.

qplot(x=a.sex, y=ahe08, data=cps.ch3, geom="boxplot")

If we wanted to see the same boxplots but separate the years we use the facets argument.

qplot(x=a.sex, y=ahe08, data=cps.ch3, geom="boxplot", facets= ~ year)

We might want to add a few things to the plot to make it easier to read, such as axis labels

qplot(x=a.sex, y=ahe08, data=cps.ch3, geom="boxplot", facets= ~ year, xlab="Sex", ylab="Average Hourly Earnings")

Another way to view the same data is to use density plots

qplot(x=ahe08, color=a.sex, data=cps.ch3, geom="density", facets= ~ year, xlab="Average Hourly Earnings")

For this part let's load another dataset, the March CPS survey for 2008 with more information on full-time workers

load("/usr/data/ec414/cps08_full.RData")

ls()

[1] "cps.08.full"   "cps.ch3"       "f"             "female.92.cps"
[5] "female.cps"    "male.cps"      "x"             "y"            
[9] "z"            

str(cps.08.full)

'data.frame':   7711 obs. of  5 variables:
 $ ahe     : num  38.46 12.5 9.86 8.24 17.79 ...
 $ year    : int  2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 ...
 $ bachelor: int  1 1 0 0 0 0 1 1 1 1 ...
 $ female  : int  0 0 0 0 0 1 0 0 0 1 ...
 $ age     : int  33 31 30 30 31 29 26 28 30 25 ...

cps.08.full$year <- factor(cps.08.full$year)
cps.08.full$bachelor <- cps.08.full$bachelor == 1
cps.08.full$female <- cps.08.full$female == 1

cps.08.bac <- subset(cps.08.full, bachelor == TRUE)

• You can find documentation for R functions in the Help menu of the RStudio environment
• A quick way to get help on a particular function is the ? command (make sure that the function's package is loaded). For example

? qplot

• Give it a try: find the documentation for the cov function
• Make use of Piazza to ask and answer questions
• Refer to the R documentation resource provided on Piazza