Introduction to ggplot2

1. Introduction
2. Dataframes
3. Different Types of Graphs:
   • Scatterplot
   • Barplot
   • Stacked bar
   • Histogram
   • Box Plot
4. Time Series/Regression
5. Other Options and Tips

ggplot2 is a unique R package used for data visualization and producing plots.

It produces fantastic-looking graphics and allows one to slice and dice data in many different ways.

Official documentation here.

It is based on the grammar of graphics:

• data set
• set of geoms
• coordinate system

To display data values:

• size
• colour
• x and y locations

Get the package:

install.packages("ggplot2")
library(ggplot2)

Consider the sample dataset:

  admit testscores  gpa rank
1     0        380 3.61    3
2     1        660 3.67    3
3     1        800 4.00    1
4     1        640 3.19    4
5     0        520 2.93    4
6     1        760 3.00    2
7     1        560 2.98    1
8     0        400 3.08    2

Selecting specific rows:

admit <- read.csv("binary.csv")
# By GPA
admit[admit$gpa>3.80,]
# By GPA & Admit
admit[admit$gpa>3.80 & admit$admit==1,]

Determine the number of students admitted with rank>=3

nrow(admit[admit$rank>=3 & admit$admit==1,])

[1] 40

Plot:

Code:

clrs <- c("black", "red", "blue", "green")[as.numeric(admit$rank)]

plot(admit$gpa, admit$testscores, xlab = "GPA", ylab = "Test Score", pch = 16, col = clrs, cex = 0.8, main = "GPA vs. Test Score")

legend("topleft", legend = c("Rank 1", "Rank 2", "Rank 3", "Rank 4"), cex = 0.8, col = c("black", "red", "blue", "green"))

library(ggplot2)

Scatterplot:

plot <- ggplot(data = admit, aes(x=gpa, y=testscores, col = rank))

plot <- plot + geom_point(size = 3)

plot <- plot + xlab("GPA") + ylab("Test Scores") + ggtitle("GPA vs. Test Scores")

Continuous Scale:

Code:

plot <- ggplot(data = admit, aes(x=gpa, y=testscores, col = factor(rank)))

plot <- plot + geom_point(size = 3) + scale_color_discrete(name = "Rank")

plot <- plot + xlab("GPA") + ylab("Test Scores") + ggtitle("GPA vs. Test Scores")

plot

Discrete Scale:

Code: Admit/No Admit Data

plot <- ggplot(data = admit, aes(x=gpa, y=testscores, col = factor(rank)))

plot <- plot + geom_point(aes(shape = factor(admit)), size = 3) + scale_color_discrete(name = "Rank") + scale_shape_discrete(name = "Admit", labels = c("No", "Yes"))

plot <- plot + xlab("GPA") + ylab("Test Scores") + ggtitle("GPA vs. Test Scores")

plot

Graph: Admit/No Admit Data

Group by rank

Group by rank

c <- ggplot(data = admit, aes(factor(rank), fill = factor(rank))) 

c <- c + geom_bar(width = 0.8) + scale_fill_discrete(name = "Rank", labels = c(1,2,3,4))

c <- c + xlab("Rank") + ylab("Frequency") + ggtitle("Group by Rank")

We've so far seen:

• scale_fill_discrete
• scale_shape_discrete
• scale_color_discrete

There are many others that take a similar format: scale_xxx_yyy

Here are some commonly-used values for xxx and yyy:

Suppose we wanted to see how many students in each rank category were admitted.

Code:

sb <- ggplot(data = admit, aes(factor(rank), fill = factor(admit))) 

sb <- sb + geom_bar(width = 0.8) + scale_fill_discrete(name = "Admit", labels = c("No", "Yes"))

sb <- sb + xlab("Rank") + ylab("Frequency") + ggtitle("Group by Rank")

sb

qplot() - Creates a complete plot with given data, geom, and mappings, Supplies many useful defaults.

qplot(x = cty, y = hwy, color = cyl, data = mpg, geom = "point") 

ggplot() - Begins a plot that you finish by adding layers to. No defaults, but provides more control than qplot().

ggplot(data = mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = cyl)) + ... + coord_flip()

Suppose we wanted to see the number of students that fall in each interval of GPA.

Code:

hist <- qplot(gpa, data = admit, geom = "histogram", binwidth = 0.1, xlab = "GPA", ylab = "Frequency", main = "GPA Ranges")

hist

ggplot() Code:

hist <- ggplot(data = admit, aes(x = gpa)) + geom_histogram(binwidth = 0.1, aes(fill = ..count..))

hist <- hist + ggtitle("GPA Ranges") + xlab("GPA") + ylab("Frequency")

hist

For gradient fill:

hist + geom_histogram(aes(fill = ..count..)) + scale_fill_gradient("Count", low = "green", high = "red")

ggplot() Plot:

Create a box plot showing the range of test scores.

Code:

boxplot <- ggplot(data = admit, aes(factor(admit), testscores)) + geom_boxplot(aes(fill = factor(admit)))

boxplot <- boxplot + ggtitle("Admission Testscores") + xlab("admit") + ylab("Test Scores") + scale_fill_discrete(name = "Admit", labels = c("No", "Yes"))

boxplot

Consider the following economics dataset:

        date   pce    pop psavert uempmed unemploy
1 1967-06-30 507.8 198712     9.8     4.5     2944
2 1967-07-31 510.9 198911     9.8     4.7     2945
3 1967-08-31 516.7 199113     9.0     4.6     2958
4 1967-09-30 513.3 199311     9.8     4.9     3143
5 1967-10-31 518.5 199498     9.7     4.7     3066
6 1967-11-30 526.2 199657     9.4     4.8     3018

Line Plots:

time_s <- ggplot(data = economics, aes(x = date, y = unemploy))

time_s <- time_s + geom_line()

time_s <- time_s + xlab("Date") + ylab("Unemployment") + ggtitle("Unemployment Time Series")

time_s

Adding some statistical transformation to this series:

We added geom_smooth()

time_s <- time_s + geom_line() + geom_smooth()

geom_smooth() - this is esssentially ggplot's built-in model fitting tool that allows you to plot the fits from any model of your choosing

• par(mfrow=c), multiple plots per page
• scaling
• removing legends
• links to other resources

In R, you can specify how many plots you want on a certain page, for instance:

par(mfrow=c(2,2))

In ggplot2, you can use the multiplot function.

multiplot(p1, p2, p3, p4, cols=2)

multiplot <- function(..., plotlist=NULL, file, cols=1, layout=NULL) {
  library(grid)

  # Make a list from the ... arguments and plotlist
  plots <- c(list(...), plotlist)

  numPlots = length(plots)

  # If layout is NULL, then use 'cols' to determine layout
  if (is.null(layout)) {
    # Make the panel
    # ncol: Number of columns of plots
    # nrow: Number of rows needed, calculated from # of cols
    layout <- matrix(seq(1, cols * ceiling(numPlots/cols)),
                    ncol = cols, nrow = ceiling(numPlots/cols))
  }

  ...
  ...
}

Controlling x-axis and y-axis limits and range:

m <- qplot(rating, votes, data=subset(movies, votes > 1000), na.rm = TRUE)

# Control x range
m <- m + scale_x_continuous(limits=c(7, 8))

# Control y range
m <- m + scale_y_continuous(limits=c(1000, 10000))

# or ...

m + xlim(7, 8)

m + ylim(1000, 10000)

Controlling x-axis and y-axis breaks: choosing where the ticks appear

# Breaks: 1 to 10
m + scale_x_continuous(breaks=1:10)

Controlling x-axis and y-axis breaks: choosing where the ticks appearand label them.

# Labelling breaks
m + scale_x_continuous(breaks=c(2,5,8), labels=c("horrible", "ok", "awesome"))

Sometimes having too many legends can overcrowd your graphs, especially when you use the multiplot function.

Code:

# Remove legend for a particular aesthetic (fill)
bp + guides(fill=FALSE)

# It can also be done when specifying the scale
bp + scale_fill_discrete(guide=FALSE)

# This removes all legends
bp + theme(legend.position="none")

• Color selection: link

• Cheat sheet: link

• Anatomy of a Plot: link

Questions?