6  Graphics

6.1 Introduction

In this chapter, we cover graphic functions in R. There are multiple ways to create graphics in R, but we will focus on the base R graphics system and the ggplot2 package from the tidyverse package. The tidyverse package consists of multiple packages that are designed for data science and data visualization. In Table 6.1, we list the core packages in the tidyverse package.

Table 6.1: The core tidyverse packages

Package	Purpose
ggplot2	Data visualization using the Grammar of Graphics
dplyr	Data manipulation (filter, summarize, mutate, arrange, etc.)
tidyr	Data tidying (reshape, pivot, separate, unite)
readr	Reading rectangular data (CSV, TSV, etc.)
tibble	Modern, user-friendly version of data frames
stringr	String manipulation
forcats	Handling categorical (factor) variables
purrr	Functional programming and iteration tools

# Load necessary libraries
library(tidyverse)
library(gridExtra)

6.2 The base R approach to graphics

The base approach to graphics consists of two types of functions: high-level plotting functions and low-level plotting functions. We use the high-level plotting functions to create a new plot and the low-level plotting functions to customize the existing plot. In Table 6.2, we list some of the commonly used high-level plotting functions in base R.

Table 6.2: High-level plotting functions

Function	Description
plot()	Scatterplot and line plot
hist()	Histogram
boxplot()	Boxplot
pie()	Pie chart
qqplot(), qqnorm(), qqline()	Quantile plots
density()	Density plot
sunflowerplot()	Sunflower scatterplot
pairs()	Scatter plot matrix
symbols()	Draw symbols on a plot
dotchart(), barplot(), pie()	Dot chart, bar chart, pie chart
curve()	Draw a curve from a given function
image()	Create a grid of colored rectangles
contour(), filled.contour()	Contour plot
persp()	Plot 3-D surface

In Table 6.3, we list some of the commonly used low-level plotting functions in base R.

Table 6.3: Low-level plotting functions

Function	Description
points()	Add points to a figure
lines()	Add lines to a figure
text()	Insert text in the plot region
mtext()	Insert text in the figure and outer margins
title()	Add figure title or outer title
legend()	Insert legend
axis(), axis.Date()	Customize axes
abline()	Add horizontal and vertical lines or a single line
box()	Draw a box around the current plot
rug()	Add a 1-D plot of the data to the figure
polygon()	Draw a polygon
rect()	Draw a rectangle
arrows()	Draw arrows
segments()	Draw line segments
trans3d()	Add 2-D components to a 3-D plot

In this part, we show how the high and low level plotting functions can be used to create different types of plots.

6.2.1 Scatterplots and line plots

We use the plot function to create scatterplots and line plots. The plot function has many arguments that allow us to customize the plot. The syntax for the plot function is as follows:

plot(x, y = NULL, type = "p", xlim = NULL, ylim = NULL, log = "", main = NULL,
     sub = NULL, xlab = NULL, ylab = NULL, ann = par("ann"), axes = TRUE,
     col = par("col"), pch = par("pch"), cex = par("cex"), lty = par("lty"),
     lwd = par("lwd"), ...)

The par function is used to set or query graphical parameters. If we type ?par in the R console, we will see a long list of graphical parameters that can be set using the par function. Below, we list some of the commonly used arguments in the plot function:

• type: 1-character string denoting the plot type
  
• xlim: x-axis limits, e.g., c(x1, x2)
  
• ylim: y-axis limits, e.g., c(y1, y2)
  
• log: Character string that contains "x" if x-axis is log-scale, "y" if y-axis is log-scale, and "xy" if both axes are log-scale
  
• main: Main title for the plot
  
• sub: Subtitle for the plot
  
• xlab: x-axis label
  
• ylab: y-axis label
  
• ann: Logical; should default annotation appear on plot
  
• axes: Logical; should both axes be drawn
  
• col: Color for lines and points; either a character string or a number that indexes the palette()
  
• pch: Number referencing a plotting symbol or a character string
  
• cex: Number giving the character expansion of the plot symbols
  
• lty: Number referencing a line type
  
• lwd: Line width

For illustrations, we use the dataset in the caschool.csv file. This dataset contains information on kindergarten through eighth-grade students across 420 California school districts in 1999. It is a district-level dataset that includes variables on average student performance and demographic characteristics. Table 6.4 provides a description of the variables in the caschool.csv file.

Table 6.4: Description of variables in the caschool dataframe

Variable	Description
dist_code	District code
read_scr	Average reading score
math_scr	Average math score
county	County
district	District
gr_span	Grade span of district
enrl_tot	Total enrollment
teachers	Number of teachers
computer	Number of computers
testscr	Average test score (= (read_scr + math_scr)/2)
comp_stu	Computers per student (= computer / enrl_tot)
expn_stu	Expenditures per student
str	Student-teacher ratio (= teachers / enrl_tot)
el_pct	Percent of English learners
meal_pct	Percent qualifying for reduced-price lunch
clw_pct	Percent qualifying for CalWorks
aving	District average income (in $1000s)

# Load the caschool dataset
caschool <- read.table("data/caschool.csv", header = TRUE, sep = ",")
# Column names of the caschool dataset
colnames(caschool)

 [1] "Observation.Number" "dist_cod"           "county"            
 [4] "district"           "gr_span"            "enrl_tot"          
 [7] "teachers"           "calw_pct"           "meal_pct"          
[10] "computer"           "testscr"            "comp_stu"          
[13] "expn_stu"           "str"                "avginc"            
[16] "el_pct"             "read_scr"           "math_scr"          

In the following code chunk, we create a scatterplot of test scores versus student-teacher ratio using the plot function.

# Scatterplot of test scores vs. student-teacher ratio
plot(caschool$str, caschool$testscr,
     type = "p", # scatterplot
     main = "Test Scores vs. Student-Teacher Ratio",
     xlab = "Student-teacher ratio",
     ylab = "Test scores",
     col = "steelblue",
     pch = 19, # solid circle
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7, # axis tick label size
     panel.first = grid() # add grid lines
     )

Figure 6.1: Scatterplot of test scores vs. student-teacher ratio

In the following code chunk, we create a new categorical variable category based on the test scores. We then use this variable to categorize schools into four groups: “Excellent”, “Good”, “Average”, and “Below Average”. We use different colors and plotting symbols to represent different categories of schools in the scatterplot. The legend is added using the legend function. The scatterplot is shown in Figure 6.2.

# Add category column based on test scores
caschool$category <- NA  # Initialize a new column for category
for (i in 1:nrow(caschool)) {
  if (caschool$testscr[i] >= 680) {
    caschool$category[i] <- "Excellent"
  } else if (caschool$testscr[i] >= 665) {
    caschool$category[i] <- "Good"
  } else if (caschool$testscr[i] >= 640) {
    caschool$category[i] <- "Average"
  } else {
    caschool$category[i] <- "Below Average"
  }
}

# Define colors and plotting symbols for each category
colors <- c("Excellent" = "darkgreen", "Good" = "blue", "Average" = "orange", "Below Average" = "red")
symbols <- c("Excellent" = 17, "Good" = 16, "Average" = 15, "Below Average" = 19)
# Scatterplot of test scores vs. income with categories
plot(caschool$str, caschool$testscr,
     type = "p", # scatterplot
     main = "Test Scores vs. Student-Teacher Ratio by Category",
     xlab = "Student-teacher ratio",
     ylab = "Test scores",
     col = colors[caschool$category],
     pch = symbols[caschool$category],
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7, # axis tick label size
     panel.first = grid() # add grid lines
     )
# Add legend
legend("topright", legend = names(colors), col = colors, pch = symbols, cex = 0.7)

Figure 6.2: Scatterplot of test scores vs. student-teacher ratio by category

In the next example, we create a line plot of average test scores versus student-teacher ratio. We first sort the dataset by student-teacher ratio by using data = caschool[order(caschool$str), ]. We then use the type = "l" argument in the plot function to create the line plot. The line plot is shown in Figure 6.3.

# Line plot of average test scores vs. student-teacher ratio
data = caschool[order(caschool$str), ]
plot(data$str, data$testscr,
     type = "l", # line plot
     main = "Average Test Scores vs. Student-Teacher Ratio",
     xlab = "Student-teacher ratio",
     ylab = "Average Test scores",
     col = "steelblue",
     lwd = 1, # line width
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7, # axis tick label size
     panel.first = grid() # add grid lines
     )

Figure 6.3: Line plot of average test scores vs. student-teacher ratio

We can also use the line plot to visualize the probability density function of a distribution. In the following example, we plot the probability density function of the standard normal distribution and the student t-distribution with 5 degrees of freedom. We use the low-level function lines to add the t-distribution curve to the existing normal distribution plot. The plot is shown in Figure 6.4.

# Probability density functions of the standard normal and t-distributions
x <- seq(-4, 4, length = 100)
y_normal <- dnorm(x, mean = 0, sd = 1)
y_t <- dt(x, df = 5)
plot(x, y_normal,
     type = "l", # line plot
     main = "Probability Density Functions",
     xlab = "x",
     ylab = "Density",
     col = "steelblue",
     lwd = 2, # line width
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7 # axis tick label size
     )
lines(x, y_t, col = "darkorange", lwd = 2)
legend("topright", legend = c("Standard Normal", "t (df=5)"), 
col = c("steelblue", "darkorange"), lwd = 2, cex = 0.7)

Figure 6.4: Probability density functions of the standard normal and t-distributions

We can also use the plot function to generate plots of probability mass functions for discrete distributions, such as the binomial and Poisson distributions. In the following example, we plot the probability mass functions of the binomial distribution with parameters size = 20 and prob = 0.75. Note that we use type = "h" to create histogram-like vertical lines and the low-level function points to add points to the plot. The plot is shown in Figure 6.5.

# Probability mass functions of the binomial distribution
x_binom <- 0:20
y_binom <- dbinom(x_binom, size = 20, prob = 0.75)
plot(x_binom, y_binom,
     type = "h", # histogram-like vertical lines
     main = "Probability Mass Function",
     xlab = "x",
     ylab = "Probability",
     col = "steelblue",
     lwd = 2, # line width
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7, # axis tick label size
     panel.first = grid()
     )
# Add points to the binomial distribution
points(x_binom, y_binom, col = "steelblue", pch = 19)

Figure 6.5: Probability mass functions of the binomial distribution

6.2.2 Histograms

We use the hist function to create histograms. The hist function has many arguments that allow us to customize the histogram. The syntax for the hist function is as follows:

hist(x, breaks = "Sturges",
     freq = NULL, probability = !freq,
     include.lowest = TRUE, right = TRUE, fuzz = 1e-7,
     density = NULL, angle = 45, col = "lightgray", border = NULL,
     main = paste("Histogram of" , xname),
     xlim = range(breaks), ylim = NULL,
     xlab = xname, ylab,
     axes = TRUE, plot = TRUE, labels = FALSE,
     nclass = NULL, warn.unused = TRUE, ...)

The breaks argument specifies the number of bins or the method to calculate the number of bins. The freq argument is a logical value that indicates whether to plot frequencies or densities. If freq = TRUE, the histogram will show frequencies; if freq = FALSE, it will show densities. In the following example, we generate a histogram of test scores. The histogram is shown in Figure 6.6.

# Histogram of test scores
hist(caschool$testscr,
     breaks = 15, # number of bins
     freq = TRUE, # plot frequencies
     col = "steelblue", # fill color
     border = "black", # border color
     main = "Histogram of Test Scores",
     xlab = "Test scores",
     ylab = "Frequency",
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7 # axis tick label size
     )

Figure 6.6: Histogram of test scores

6.2.3 Boxplots

A boxplot can be created using the boxplot function. Its syntax is as follows:

boxplot(formula = NULL, data = NULL,
        subset = NULL, na.action = na.omit,
        main = NULL, xlab = NULL, ylab = NULL,
        border = par("fg"), col = NULL,
        names = NULL, at = NULL,
        horizontal = FALSE, notch = FALSE,
        varwidth = FALSE, outline = TRUE,
        pars = list(), ...)

The important argument is formula, which specifies the response variable and the grouping variable. The syntax for the formula argument is response ~ group, where response is the numeric variable to be plotted and group is the categorical variable that defines the groups. In the following example, we create a boxplot of test scores by category. The boxplot is shown in Figure 6.7.

# Boxplot of test scores by category
boxplot(testscr ~ category, data = caschool,
        main = "Boxplot of Test Scores by Category",
        xlab = "Category",
        ylab = "Test scores",
        col = c("darkgreen", "blue", "orange", "red"),
        cex.lab = 0.8, # axis label size
        cex.main = 0.9, # main title size
        cex.axis = 0.7 # axis tick label size
        )

Figure 6.7: Boxplot of test scores by category

6.2.4 Pie charts

We use the pie function to create pie charts. Its syntax is as follows:

pie(x, labels = NULL, edges = 200, radius = 0.8,
    col = NULL, border = "white", lty = NULL, lwd = NULL,
    main = NULL, sub = NULL, cex.main = 1, cex.sub = 1,
    cex = 1, ...)

The important argument is x, which is a numeric vector that contains the values to be plotted. In the following example, we create a pie chart of the number of schools in each category. The pie chart is shown in Figure 6.8.

# Pie chart of number of schools by category
category_counts <- table(caschool$category)
pie(category_counts,
    labels = names(category_counts),
    col = c("darkgreen", "blue", "orange", "red"),
    main = "Pie Chart of Number of Schools by Category",
    cex.main = 0.9 # main title size
    )

Figure 6.8: Pie chart of number of schools by category

6.2.5 Quantile plots

We use the qqplot, qqnorm, and qqline functions to create quantile plots. The qqnorm function creates a normal Q-Q plot, and the qqline function adds a reference line to the plot. In the following example, we create the normal Q-Q plot of test scores.

# Normal Q-Q plot of test scores
qqnorm(caschool$testscr,
        main = "Normal Q-Q Plot of Test Scores",
        col = "steelblue",
        pch = 1, # hollow circle
        cex.lab = 0.8, # axis label size
        cex.main = 0.9, # main title size
        cex.axis = 0.7 # axis tick label size
        )
qqline(caschool$testscr, col = "red", lwd = 2)

Figure 6.9: Normal Q-Q plot of test scores

6.2.6 Density plots

The density function estimates the kernel density of a continuous variable. In the following example, we create the density plot of test scores. We use the low-level function polygon to fill the area under the density curve. The density plot is shown in Figure 6.10.

# Density plot of test scores
density_testscr <- density(caschool$testscr)
plot(density_testscr,
     main = "Density Plot of Test Scores",
     xlab = "Test scores",
     ylab = "Density",
     col = "steelblue",
     lwd = 2, # line width
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7 # axis tick label size
     )
# Fill the area under the density curve
polygon(density_testscr, col = "lightblue", border = "black") 

Figure 6.10: Density plot of test scores

In the following example, we show the density plot and histogram of test scores together in one figure. Note that we set the freq = FALSE argument in the hist function to plot densities instead of frequencies. Also, we use the low-level function lines to add the density curve to the histogram. The figure is shown in Figure 6.11.

# Density plot and histogram of test scores
hist(caschool$testscr,
     breaks = 15, # number of bins
     freq = FALSE, # plot densities
     col = "lightgray", # fill color
     border = "black", # border color
     main = "Density Plot and Histogram of Test Scores",
     xlab = "Test scores",
     ylab = "Density",
     cex.lab = 0.8, # axis label size
     cex.main = 0.9, # main title size
     cex.axis = 0.7 # axis tick label size
     )
lines(density(caschool$testscr), col = "steelblue", lwd = 2)

Figure 6.11: Density plot and histogram of test scores

6.2.7 Multiple plots in one figure

We can use the par function to create multiple plots in one figure. The mfrow argument in the par function specifies the number of rows and columns of plots in the figure. In the following example, we create a 2x2 grid of plots: a scatterplot, a histogram, a boxplot, and a density plot. The figure is shown in Figure 6.12.

# Multiple plots in one figure
par(mfrow = c(2, 2)) # 2 rows and 2 columns
# Scatterplot
plot(caschool$str, caschool$testscr,
     main = "Scatterplot of Test Scores vs. Student-Teacher Ratio",
     xlab = "Student-teacher ratio",
     ylab = "Test scores",
     col = "steelblue",
     pch = 19
     )
# Histogram
hist(caschool$testscr,
        breaks = 15,
        freq = TRUE,
        col = "steelblue",
        border = "black",
        main = "Histogram of Test Scores",
        xlab = "Test scores",
        ylab = "Frequency"
        )
# Boxplot
boxplot(testscr ~ category, data = caschool,
        main = "Boxplot of Test Scores by Category",
        xlab = "Category",
        ylab = "Test scores",
        col = c("darkgreen", "blue", "orange", "red")
        )
# Density plot
density_testscr <- density(caschool$testscr)
plot(density_testscr,
     main = "Density Plot of Test Scores",
     xlab = "Test scores",
     ylab = "Density",
     col = "steelblue",
     lwd = 2
     )
polygon(density_testscr, col = "lightblue", border = "black")
# Reset par to default
par(mfrow = c(1, 1))

Figure 6.12: Multiple plots in one figure

6.3 The tidyverse approach to graphics

The ggplot2 package in the tidyverse package provides a flexible way to create graphics based on the Grammar of Graphics. In this part, we cover the basic concepts and functions in ggplot2 for creating graphics.

6.3.1 Basic concepts in ggplot2

The basic idea of ggplot2 is to build a plot layer by layer. The ggplot function creates a ggplot object, which is the foundation of the plot. Its syntax is as follows:

ggplot(data = NULL, mapping = aes(), ..., environment = parent.frame())

where

• data: A data frame containing the variables to be plotted.
• mapping: Aesthetic mappings created by the aes function. This argument defines how variables in the data are mapped to visual properties (aesthetics) of the plot, such as x and y coordinates, colors, shapes, and sizes. We use the aes function to create aesthetic mappings.
• ...: Additional arguments passed to the ggplot function.

Below, we create a ggplot object using the caschool dataset. The aes function is used to specify the x and y aesthetics, which map the str variable to the x-axis and the testscr variable to the y-axis.

# Create a ggplot object
ggplot(data = caschool, aes(x = str, y = testscr))

Figure 6.13: Creating a ggplot object

We then add layers to the ggplot object using the + operator. Layers are created using geom functions, such as geom_point, geom_line, geom_histogram, geom_boxplot, and geom_density. Basically, geoms correspond to the high-level plotting functions in the base approach and are the geometric objects that represent the data points in the plot. In Table 6.5, we list some commonly used geoms in ggplot2.

Table 6.5: Commonly used geoms in ggplot2

Geom	Description
Graphical primitives
geom_blank()	Display nothing. Most useful for adjusting axes limits.
geom_point()	Points.
geom_path()	Paths.
geom_ribbon()	Ribbons, a path with vertical thickness.
geom_segment()	A line segment, specified by start and end position.
geom_rect()	Rectangles.
geom_polygon()	Filled polygons.
geom_text()	Text.
One variable
geom_bar()	Display distribution of discrete variable.
geom_histogram()	Bin and count continuous variable, display with bars.
geom_density()	Smoothed density estimate.
geom_dotplot()	Stack individual points into a dot plot.
geom_freqpoly()	Bin and count continuous variable, display with lines.
Two variables
geom_point()	Scatterplot.
geom_quantile()	Smoothed quantile regression.
geom_rug()	Marginal rug plots.
geom_smooth()	Smoothed line of best fit.
geom_text()	Text labels.
geom_bin2d()	Bin into rectangles and count.
geom_density2d()	Smoothed 2D density estimate.
geom_hex()	Bin into hexagons and count.
geom_count()	Count number of points at distinct locations.
geom_jitter()	Randomly jitter overlapping points.
geom_bar(stat = "identity")	A bar chart of precomputed summaries.
geom_boxplot()	Boxplots.
geom_violin()	Show density of values in each group.
geom_area()	Area plot.
geom_line()	Line plot.
geom_step()	Step plot.
geom_crossbar()	Vertical bar with center.
geom_errorbar()	Error bars.
geom_linerange()	Vertical line.
geom_map()	Fast version of geom_polygon() for map data.
Three variables
geom_contour()	Contours.
geom_tile()	Tile the plane with rectangles.
geom_raster()	Fast version of geom_tile() for equal-sized tiles.
geom_contour_filled()	Filled contours.

Each geom function understands a set of aesthetic mappings that define how variables in the data are mapped to visual properties of the geom. For example, the geom_point function understands the following aesthetics (type ?geom_point in the R console):

• x: x-coordinate of the points
  
• y: y-coordinate of the points
• color: color of the points
• size: size of the points
• shape: shape of the points
• alpha: transparency of the points
• fill: fill color of the points (for shapes that have a fill)
• stroke: width of the point border (for shapes that have a border)
• group: grouping variable for the points

We can either use aesthetics in the aes function or use them as arguments in the geom_point function. If we use an aesthetic inside the aes function, it is mapped to a variable in the data. However, if we use an aesthetic outside the aes function, it is set to a constant value, allowing us to customize the appearance of the geom.

6.3.2 Scatterplots

In the following example, we create a scatterplot of test scores versus student-teacher ratio. The color and size of the points are specified using the color and size arguments in the geom_point function. The labs function is used to add titles and labels to the plot. The scatterplot is shown in Figure 6.14.

# Scatterplot of test scores vs. student-teacher ratio
ggplot(data = caschool, aes(x = str, y = testscr)) +
  geom_point(color = "steelblue", size = 2) +
  labs(title = "Test Scores vs. Student-Teacher Ratio",
       x = "Student-teacher ratio",
       y = "Test scores")

Figure 6.14: Scatterplot of test scores vs. student-teacher ratio

In the above example, we set the color and size aesthetics to constant values in the geom_point function. We can also map these aesthetics to variables in the data using the aes function, as shown below.

In the following example, we map the color aesthetic to the category variable in the aes function. This automatically assigns different colors to the points based on their category. The scatterplot is shown in Figure 6.15.

# Scatterplot of test scores vs. student-teacher ratio by category
ggplot(data = caschool, aes(x = str, y = testscr, color = category)) +
  geom_point(size = 2) +
  labs(title = "Test Scores vs. Student-Teacher Ratio by Category",
       x = "Student-teacher ratio",
       y = "Test scores") 

Figure 6.15: Scatterplot of test scores vs. student-teacher ratio by category

We can change the legend title and labels using the scale_color_manual function. In the following example, we customize the colors and legend of the scatterplot. The scatterplot is shown in Figure 6.16.

# Customized scatterplot of test scores vs. student-teacher ratio by category 
# Convert category to factor with specified levels
caschool$category <- factor(caschool$category, 
                            levels = c("Excellent", "Good", "Average", "Below Average"))

# Define custom colors
values <- c("Excellent" = "darkgreen", "Good" = "blue", 
            "Average" = "orange", "Below Average" = "red")

# Define custom labels
labels <- c("Excellent", "Good", "Average", "Below Average")

# Scatterplot 
ggplot(data = caschool, aes(x = str, y = testscr, color = category)) +
  geom_point(size = 2) +
  scale_color_manual(values = values,
                     name = "School Category",
                     labels = labels) +
  labs(title = "Test Scores vs. Student-Teacher Ratio by Category",
       x = "Student-teacher ratio",
       y = "Test scores")

Figure 6.16: Customized scatterplot of test scores vs. student-teacher ratio by category

In the following example, we first create a binary variable str_dummy that indicates whether the student-teacher ratio is above or below 20. We then use the geom_jitter function to create a scatterplot of testscr versus str_dummy with jittered points. The geom_jitter function adds random noise to the points to reduce overplotting. The scatterplot is shown in Figure 6.17.

# Jittered scatterplot of test scores vs. str_dummy
caschool$str_dummy <- ifelse(caschool$str > 20, "Above 20", "20 or Below")
ggplot(data = caschool, aes(x = str_dummy, y = testscr, color = category)) +
  geom_jitter(width = 0.1, height = 0, size = 2, shape = 16) +
  labs(x = "Student-teacher ratio",
       y = "Test scores")

Figure 6.17: Jittered scatterplot of testscr vs. str_dummy

6.3.3 Line plots

In the following example, we create a line plot of average test scores versus student-teacher ratio. Recall that we created a new dataset data that is sorted by student-teacher ratio. We use the geom_line function to create the line plot. The line plot is shown in Figure 6.18.

# Line plot of average test scores vs. student-teacher ratio
data = caschool[order(caschool$str), ]
ggplot(data = data, aes(x = str, y = testscr)) +
  geom_line(color = "steelblue", linewidth = 1) +
  labs(title = "Average Test Scores vs. Student-Teacher Ratio",
       x = "Student-teacher ratio",
       y = "Average Test scores")

Figure 6.18: Line plot of average test scores vs. student-teacher ratio

We can also use the geom_line function to plot the probability density function of a distribution. In the following example, we plot the probability density function of the standard normal distribution and the student t-distribution with 5 degrees of freedom. We use two geom_line functions to add the two curves to the plot. We use the color = "Standard Normal" and color = "t (df=5)" arguments inside the aes function to map the colors of the lines to the legend. We then use the scale_color_manual function to customize the colors and legend title. The plot is shown in Figure 6.19.

# Probability density functions of the standard normal and t-distributions
x <- seq(-4, 4, length = 100)
y_normal <- dnorm(x, mean = 0, sd = 1)
y_t <- dt(x, df = 5)
df <- data.frame(x = x, y_normal = y_normal, y_t = y_t)
ggplot(data = df, aes(x = x)) +
  geom_line(aes(y = y_normal, color = "Standard Normal"), linewidth = 1) +
  geom_line(aes(y = y_t, color = "t (df=5)"), linewidth = 1) +
  scale_color_manual(
     name = "Distribution",
     values = c("Standard Normal" = "steelblue", "t (df=5)" = "darkorange")) +
  labs(x = "x", y = "Density")

Figure 6.19: Probability density functions of the standard normal and t-distributions

The ggplot version of the probability mass function of the binomial distribution shown in Figure 6.5 can be generated using the geom_segment and geom_point functions. The plot is shown in Figure 6.20.

# Probability mass functions of the binomial distribution
# Data
x_binom <- 0:20
y_binom <- dbinom(x_binom, size = 20, prob = 0.75)
df_binom <- data.frame(x = x_binom, Probability = y_binom)

# Plot
ggplot(data = df_binom, aes(x = x, y = Probability)) +
  geom_segment(aes(xend = x, y = 0, yend = Probability), color = "steelblue", linewidth = 1.2) +
  geom_point(color = "steelblue", size = 3) +
  labs(
    title = "Probability Mass Function",
    x = "x",
    y = "Probability"
  )

Figure 6.20: Probability mass functions of the binomial distribution

The geom_smooth function can be used to add a smoothed line of best fit to a scatterplot. In the following example, we create a scatterplot of test scores versus student-teacher ratio and add a smoothed line of best fit using the geom_smooth function. The plot is shown in Figure 6.21.

# Scatterplot of test scores vs. student-teacher ratio with smoothed line
ggplot(data = caschool, aes(x = str, y = testscr)) +
  geom_point(color = "steelblue", size = 2) +
  geom_smooth(method = "loess", color = "darkorange", se = TRUE) +
  labs(x = "Student-teacher ratio",
       y = "Test scores")

Figure 6.21: Scatterplot of test scores vs. student-teacher ratio with smoothed line

6.3.4 Histograms

The syntax for the geom_histogram function is as follows:

geom_histogram(mapping = NULL, data = NULL,
               stat = "bin", position = "stack",
               ...,
               binwidth = NULL, bins = NULL,
               na.rm = FALSE, show.legend = NA,
               inherit.aes = TRUE)

It understands the following aesthetics: (i) x: x-coordinate of the bars, (ii) y: height of the bars (computed by default), (iii) fill: fill color of the bars, and (iv) color: border color of the bars.

In the following example, we generate the histogram of test scores. The argument binwidth = 5 specifies the width of each bin. The histogram is shown in Figure 6.22.

# Histogram of test scores
ggplot(data = caschool, aes(x = testscr)) +
  geom_histogram(binwidth = 5, fill = "steelblue", color = "white") +
  labs(x = "Test scores",
       y = "Count")

Figure 6.22: Histogram of test scores

6.3.5 Bar charts

The geom_bar function is similar to the geom_histogram function, but it is used for categorical variables. It also assumes the same aesthetics as the geom_histogram function. In the following example, we generate a bar chart of the number of school districts in each category. The bar chart is shown in Figure 6.23.

# Bar chart of students by category
ggplot(data = caschool, aes(x = category)) +
  geom_bar(fill = "steelblue", color = "white") +
  labs(x = "Category",
       y = "Count")

Figure 6.23: Bar chart of students by category

We can also use the geom_bar function to create pie charts. In the following example, we create a pie chart of the number of school districts in each category. Here, the option width = 1 in the geom_bar function makes the bars full width, and the coord_polar function transforms the bar chart into a pie chart. The theme_void function removes the axes and background for a cleaner look. The pie chart is shown in Figure 6.24.

# Pie chart of number of schools by category
ggplot(data = caschool, aes(x = "", fill = category)) +
     geom_bar(width = 1) +
     coord_polar(theta = "y") +
     scale_fill_manual(values = c("red", "darkgreen", "orange", "steelblue"),
                       name = "Category") +
     labs(title = "Number of Schools by Category") +
     theme_void()

Figure 6.24: Pie chart of number of schools by category

6.3.6 Boxplots

We use the geom_boxplot function to create boxplots. It assumes the same aesthetics as the geom_bar function. Its syntax is as follows:

geom_boxplot(mapping = NULL, data = NULL,
                stat = "boxplot", position = "dodge",
                ...,
                outlier.colour = NULL,
                outlier.shape = 19,
                outlier.size = 1.5,
                outlier.stroke = 0.5,
                notch = FALSE,
                na.rm = FALSE,
                show.legend = NA,
                inherit.aes = TRUE)

In the following example, we create a boxplot of test scores by category. Here, we use the fill aesthetic to specify different fill colors for each category. To hide the legend, we use theme(legend.position = "none"). The boxplot is shown in Figure 6.25.

# Boxplot of test scores by category
ggplot(data = caschool, aes(x = category, y = testscr, fill = category)) +
  geom_boxplot() +
  scale_fill_manual(values = c("darkgreen", "blue", "orange", "magenta")) +
  labs(title = "Boxplot of Test Scores by Category",
       x = "Category",
       y = "Test scores") +
  theme(legend.position = "none")

Figure 6.25: Boxplot of test scores by category

In the following example, we do not use the fill aesthetic inside the aes function. Instead, we set the fill argument in the geom_boxplot function to a constant value. This creates a boxplot with the same fill color for all categories. If we want to assign a different color to each box, we can use fill = c("darkgreen", "blue", "orange", "magenta") in the geom_boxplot function. The boxplot is shown in Figure 6.26.

# Boxplot of test scores by category with constant fill color
ggplot(data = caschool, aes(x = category, y = testscr)) +
  geom_boxplot(fill = "steelblue") +
  labs(title = "Boxplot of Test Scores by Category",
       x = "Category",
       y = "Test scores")

Figure 6.26: Boxplot of test scores by category with constant fill color

6.3.7 Quantile plots

The geom_qq function creates the normal Q-Q plot, and the geom_qq_line function adds a reference line to the plot. In the following example, we create the normal Q-Q plot of test scores.

# Normal Q-Q plot of test scores
ggplot(data = caschool, aes(sample = testscr)) +
  geom_qq(color = "steelblue", size = 2, shape = 19) +
  geom_qq_line(color = "red", linewidth = 1) +
  labs(title = "Normal Q-Q Plot of Test Scores",
       x = "Theoretical Quantiles",
       y = "Sample Quantiles")

Figure 6.27: Normal Q-Q plot of test scores

6.3.8 Density plots

The geom_density function can be used to create density plots. Its syntax is as follows:

geom_density(mapping = NULL, data = NULL,
                 stat = "density", position = "identity",
                 ...,
                 na.rm = FALSE, show.legend = NA,
                 inherit.aes = TRUE)

It understands the following aesthetics: (i) x: x-coordinate of the density curve, (ii) y: height of the density curve (computed by default), (iii) fill: fill color of the area under the density curve, (iv) color: border color of the density curve, (v) linetype: line type of the density curve, and (vi) size: line width of the density curve, (vii) alpha: transparency of the fill color, and (viii) group: grouping variable for the density curve.

In the following example, we create a density plot of test scores.

# Density plot of test scores
ggplot(data = caschool, aes(x = testscr)) +
     geom_density(fill = "lightblue", color = "black", linewidth = 1) +
     labs(title = "Density Plot of Test Scores",
           x = "Test scores",
           y = "Density")

Figure 6.28: Density plot of test scores

We can overlay a density plot on a histogram. In the following example, we create a histogram of test scores and overlay a density plot on it. The plot is shown in Figure 6.29.

# Density plot and histogram of test scores
ggplot(data = caschool, aes(x = testscr)) +
     geom_histogram(aes(y = after_stat(density)), binwidth = 5,
                    fill = "lightgray", color = "black") +
     geom_density(color = "steelblue", linewidth = 1) +
     labs(title = "Density Plot and Histogram of Test Scores",
          x = "Test scores",
          y = "Density")

Figure 6.29: Density plot and histogram of test scores

6.3.9 Multiple plots in one figure

We can use the gridExtra package to create multiple plots in one figure. The grid.arrange function in the gridExtra package arranges multiple ggplot objects in a grid layout. In the following example, we create a 2x2 grid of plots. The figure is shown in Figure 6.30.

# Multiple plots in one figure
# Scatterplot
p1 <- ggplot(data = caschool, aes(x = str, y = testscr)) +
        geom_point(color = "steelblue", size = 2) +
        labs(title = "Scatterplot of Test Scores vs. Student-Teacher Ratio",
             x = "Student-teacher ratio",
             y = "Test scores")
# Histogram
p2 <- ggplot(data = caschool, aes(x = testscr)) +
        geom_histogram(binwidth = 5, fill = "steelblue", color = "white") +
        labs(title = "Histogram of Test Scores",
             x = "Test scores",
             y = "Count")
# Boxplot
p3 <- ggplot(data = caschool, aes(x = category, y = testscr
)) +
        geom_boxplot(fill = c("darkgreen", "blue", "orange", "red")) +
        labs(title = "Boxplot of Test Scores by Category",
             x = "Category",
             y = "Test scores")
# Density plot
p4 <- ggplot(data = caschool, aes(x = testscr)) +
        geom_density(fill = "lightblue", color = "black", linewidth = 1) +
        labs(title = "Density Plot of Test Scores",
             x = "Test scores",
             y = "Density")
# Arrange plots in a 2x2 grid
grid.arrange(p1, p2, p3, p4, nrow = 2, ncol = 2)

Figure 6.30: Multiple plots in one figure

6.3.10 Themes in ggplot2

In ggplot2, there are several built-in themes to customize the appearance of plots. In Table 6.6, we list some commonly used themes.

Table 6.6: Commonly used themes in ggplot2

Theme	Description
theme_gray()	Default ggplot2 theme with gray background and white grid lines.
theme_bw()	Black and white theme with white background and black grid lines.
theme_minimal()	Minimal theme with no background and grid lines.
theme_classic()	Classic theme with white background and no grid lines.
theme_light()	Light theme with light gray background and white grid lines.
theme_dark()	Dark theme with dark gray background and light grid lines.
theme_void()	Theme with no background, grid lines, or axes.

In the following example, we create a scatterplot of test scores versus student-teacher ratio using the theme_classical theme. The scatterplot is shown in Figure 6.31.

# Scatterplot of test scores vs. student-teacher ratio with minimal theme
ggplot(data = caschool, aes(x = str, y = testscr)) +
  geom_point(color = "steelblue", size = 2) +
  labs(title = "Test Scores vs. Student-Teacher Ratio",
       x = "Student-teacher ratio",
       y = "Test scores") +
  theme_classic()

Figure 6.31: Scatterplot of test scores vs. student-teacher ratio with minimal theme

6.4 Further reading

For further reading on R graphics, we recommend Jones, Harden, and Crawley (2022) for the base R graphics system, and Wickham (2016) and Wickham, Grolemund, and Çetinkaya-Rundel (2023) for the ggplot2 package.

Jones, Elinor, Simon Harden, and Michael J. Crawley. 2022. The r Book. 2nd ed. Hoboken, NJ: Wiley. https://books.google.com.tr/books?id=xAOEEAAAQBAJ.

Wickham, Hadley. 2016. Ggplot2: Elegant Graphics for Data Analysis. 2nd ed. Use r! New York: Springer. https://ggplot2-book.org/.

Wickham, Hadley, Garrett Grolemund, and Mine Çetinkaya-Rundel. 2023. R for Data Science: Import, Tidy, Transform, Visualize, and Model Data. 2nd ed. Sebastopol, CA: O’Reilly Media, Inc. https://r4ds.had.co.nz/.