In olobiolo/Rdlazer: Resources for Teaching R

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.height = 2.5,
  fig.width = 5
)

Introduction To ggplot2

The ggplot package introduced a new standard in generating plots in R.

ggplot was later superseded by ggplot2 which introduced a special syntax that made building plots much more convenient. ggplot2 is now the gold standard of plotting data in R but naturally it is not the be all and end all. Other packages and more will likely come.

ggplot2 is now considered a part of the Tidyverse, even though it was created before. Nevertheless, the two go together well. ggplot2 "likes" tidy data and is easily added to a pipe stream.

Plots are objects so they can be drawn immediately or assigned to a variable. This way they can be held on to or modified down the line. The plot is drawn when the object is printed. In an interactive session this is done whenever the plot is put into the console but in a programming context an explicit call to print is necessary. Keep this in mind, it can be quite surprising.

creating a plot

library(ggplot2)

A plot object has a list structure and is constructed in steps.

1. Initiate a plot with a call to ggplot.

ggplot()

1. Specify the data frame from which information will be plotted in the data argument. The data becomes part of the plot object but still nothing is drawn.

ggplot(data = iris)

1. Define mappings, i.e. assign variables in the data to plot features, e.g. axes or colors. This is done by calling the aes function and enumerating the bindings, using non standard evaluation. aes_string still exists for passing column names as character strings, which is useful in programming, but it is being phased out.
   Once variables are mapped to the axes, the scales and limits are set and a grid is created.

ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Species))

1. Add geometries, actually drawn plot elements. Every geometry is created by its own function, e.g. points are drawn with geom_point and boxplots with geom_boxplot. Geometries are added with a special operator: +.

ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Species)) +
  geom_point()
# note that a legend is created automatically

| Geometry functions have data and mappings arguments like ggplot but they can inherit these arguments from the ggplot call that initiated the plot. Every geometry has a set of aesthetics it can accept to customize its looks according to the data.

| The appearance of a geometry can be modified outside of the mappings. Some geometries have certain requirements, e.g. geom_boxplot expects a discrete variable mapped to one of the axes.

ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Species)) + geom_point(pch = 1)

1. Multiple geometries can be added to one plot, where they stack on one another.

ggplot(iris, aes(x = Species, y = Sepal.Length)) +
  geom_boxplot() +
  geom_point(aes(color = Species), position = position_jitter(width = 0.2))

1. Any change to the plot is made by adding another element to the plot like a geometry. The list of things that can be done is quite long:
   • ggtitle: add a title with
   • xlab and ylab: change axis labels
   • xlim and ylim: change axes limits
   • scale_x_log10 and scale_y_log10: change axes to log scale
   • scale_color_manual: change the automatic colors for the color aesthetic
   • theme: change any plot element | The theme function can customize any element of the plot. Sets of customizations can be saved as a separate theme and applied to a plot in bulk. There are a few themes built in like theme_linedraw or theme_minimal. More can be downloaded or created.

modifying a plot

At any stage of construction a plot object can be assigned to a variable, which can make customization easier.

p <- ggplot(iris, aes(Sepal.Length, Petal.Length, color = Species))

p +geom_point()
p +geom_point() +geom_smooth()

splitting a plot

A plot can be broken up into panels according to one or more discrete variables. This is done with facet_wrap and facet_grid. Both functions accept formulas for easy specification of the facetting.

p +geom_point() +geom_smooth() +facet_grid(. ~ Species)
p +geom_point() +geom_smooth() +facet_grid(Species ~ .)

These are not separate plots! They are panels that contain subsets of the data.

multiple pltos

Winston Chang has developed a very nice function for putting several ggplot objects on one page. His original post, along with the code, can be found here. I have taken the liberty of adding the function to my utilities package acutils which can be found on Github.

exporting plots

To save a plot to a file open a graphics device, print a plot, and close the device, just as usual. An alternative is to use the ggsave function. It takes the last created ggplot object and sends it to a graphical device, which it automatically closes afterwards.

Unfortunately, ggsave will not capture a multiplot, only the last panel. multiplots must be exported normally.

Further Reading

This is just a primer into ggplot2, the subject is far too vast to cover here. There are excellent resources on the Internet anyway. Check out the links below and learn to google solutions. Good luck!

official RStudio cheat sheet
Graphs book
ZevRoss
STDHA

olobiolo/Rdlazer documentation built on Aug. 6, 2022, 11:37 a.m.