Graphical visualizations in R using ggplot2

R For SAS Users

Melinda Higgins, PhD

Research Professor/Senior Biostatistician Emory University

ggplot2 package

ggplot2 hexsticker logo

ggplot2 is a powerful graphics package for R
"GG" in ggplot stands for the "grammar of graphics"
ggplot2 uses a layering approach to build graphics
One or more geometric objects are added to the base graphics layer

Layers - base layer

# Create plot for x=sex and y=diameter
ggplot(data = abalone, aes(sex, diameter))

Define base layer with ggplot()
Set data = abalone
Set aes to sex and diameter
No graphical objects in the plot yet
x-axis is ready for sex
y-axis is ready for diameter
grid is laid out

ggplot2 graphic environment no boxplot yet

Layers - add boxplot geom

# Add boxplot geometric object or geom
ggplot(data = abalone,
       aes(sex, diameter)) +
  geom_boxplot()

Plus operator + adds layer
Boxplot geom_boxplot() added
Result is series of boxplots
Abalone diameters by sex
F females, I infants, and M males

ggplot2 boxplots added

Layers - add a theme

# Add black white theme
ggplot(data = abalone,
       aes(sex, diameter)) +
  geom_boxplot() +
  theme_bw()

Add "theme" layer using theme_bw()
Removes grey background
Draws black box around the plot

ggplot2 boxplot black white theme added

Change boxplot geom to violin geom

# Change to geom_violin()
ggplot(data = abalone,
       aes(sex, diameter)) +
  geom_violin() +
  theme_bw()

geom_violin replaces geom_boxplot
Creates a shape similar to a violin
Reflects data density distribution
Simple change to make new figure

ggplot2 violin plot

Single variable histogram

# Make histogram of shuckedWeight
ggplot(abalone, aes(shuckedWeight)) +
  geom_histogram()

Create a histogram for one variable
One variable = one aesthetic
Add geom_histogram()
Set aes() to shuckedWeight
Default colors need changing

ggplot2 histogram all black

Histogram add colors

# Make lines black and fill light blue
ggplot(abalone, aes(shuckedWeight)) +
  geom_histogram(color = "black",
                 fill = "lightblue")

Change graphical parameters
Set color of bin lines
Set fill color for bins
Each option is set inside the ()
Histogram looks much better

ggplot2 histogram with blue bars and black lines around bars

Histogram add title and axis labels

# Add x, y axis labels and title
ggplot(abalone, aes(shuckedWeight)) +
  geom_histogram(color = "black",
                 fill = "lightblue") +
  xlab("Shucked Weight") +
  ylab("Frequency Counts") +
  ggtitle("Shucked Weights Histogram")

Add better labels for axes and title
Use xlab() and ylab() for axes
Use ggtitle() for title
This figure is ready to publish!

ggplot2 hitogram axis labels and title added

Make scatterplot

# Make scatterplot with geom_point()
ggplot(abalone,
       aes(rings, shellWeight)) +
  geom_point()

A scatterplot aes needs two variables
geom_point() adds the points
Scatterplot of shell weights by rings

ggplot2 scatterplot of shellweight by rings

Scatterplot add smoothed fit line

# Add smoothed fit line
ggplot(abalone,
       aes(rings, shellWeight)) +
  geom_point() +
  geom_smooth()

To scatterplot, add geom_smooth() line
Includes shaded confidence area

ggplot2 scatterplot with smoothed fit line added

Create panels by another variable

# Add panels using facet_wrap()
ggplot(abalone,
       aes(rings, shellWeight)) +
  geom_point() +
  geom_smooth() +
  facet_wrap(vars(sex))

Add one more layer to scatterplot
Create panels for each abalone sex
Add facet_wrap() layer
vars(sex) defines variable for panels

ggplot2 scatterplot with 3 panels by sex

Rest of course

Chapter 1 finishes with brief introduction to graphics
- ggplot2 graphical skills foundation
- visualize abalone measurements by sex
Chapter 2 teaches data wrangling skills
- clean up abalone dataset
Chapter 3 teaches data exploration methods
- descriptive statistics, correlations and comparison tests
Chapter 4 teaches modeling and results presentation
- predict abalone ages by measurements
- explore models by sex

Let's make some plots for abalones

R For SAS Users