Introduction to ggplot2 theme()

June Choe

Overview

This is a short(!) introduction to the design principles of theme, to help you:

  • Get a feel for what parts of the plot can be modified how

  • Develop a vocabulary for specific components of a plot

  • Explore theme extensions developed by the community


Slides: yjunechoe.github.io/ggplot2-theme-intro

What is Theme in ggplot?

Broadly, it’s the style of your plot. But more technically, Theme refers to styles that are not mapped from the data

Is a Theme:

  • The font of the plot’s title

  • The color of the x-axis line

Is NOT a Theme:

Data

library(tidyverse)
yeah_mhm <- read_csv("yeah_mhm-small.csv")
yeah_mhm
# A tibble: 66 × 3
   participant textLower stanceType   
   <chr>       <chr>     <chr>        
 1 WOOD        mhm       affirmation  
 2 WOOD        mhm       other        
 3 ALICE       yeah      other        
 4 ALICE       yeah      other        
 5 MARY        yeah      understanding
 6 MARY        yeah      understanding
 7 MARY        yeah      affirmation  
 8 ALICE       mhm       affirmation  
 9 MARY        mhm       understanding
10 MARY        mhm       understanding
# ℹ 56 more rows

Plot

bar_plot <- ggplot(yeah_mhm, aes(x = stanceType)) +
  geom_bar(aes(fill = textLower))
bar_plot

Theme in code

“Color the x-axis line red”:

bar_plot +
  theme(axis.line.x = element_line(color = "red"))

The structure of theme()

theme( COMPONENT = STYLE )

“The x-axis line should be a red colored line”:

theme(axis.line.x = element_line(color = "red"))


Components are hierarchically organized pieces of a plot

Styles are values specifying the appearance of a component

Theme families

theme() has lots of options, but not that scary to navigate!

Theme elements are organized hierarchically:

theme(
  axis.line,
  axis.line.x,
  axis.line.x.top,
  axis.line.x.bottom,
  axis.line.y,
  axis.line.y.left,
  axis.line.y.right,
)

Theme families

theme() has lots of options, but not that scary to navigate!

Theme elements are organized hierarchically:

theme(
  axis.line,
    axis.line.x,
      axis.line.x.top,
      axis.line.x.bottom,
    axis.line.y,
      axis.line.y.left,
      axis.line.y.right,
)

Styles are inherited if left unspecified

Theme inheritance: axis line

“X-axis line is red”

bar_plot +
  theme(axis.line.x = element_line(color = "red"))

Theme inheritance: axis line

“Axis lines are thick, x-axis line is red”

bar_plot +
  theme(
    axis.line   = element_line(linewidth = 2),
    axis.line.x = element_line(color = "red")
  )

Theme inheritance: axis title

theme(
  axis.title,
    axis.title.x,
      axis.title.x.top,
      axis.title.x.bottom,
    axis.title.y,
      axis.title.y.left,
      axis.title.y.right,
)

Theme inheritance: axis title

“Axis titles are big”

bar_plot +
  theme(
    axis.title = element_text(size = 20)
  )

Theme inheritance: axis title

“Axis titles are big, x-axis title is blue”

bar_plot +
  theme(
    axis.title   = element_text(size = 20),
    axis.title.x = element_text(color = "blue")
  )

Theme inheritance: axis title

“Text is in italics, axis titles are big, x-axis title is blue”

bar_plot +
  theme(
    text         = element_text(face = "italic"),
    axis.title   = element_text(size = 20),
    axis.title.x = element_text(color = "blue")
  )

Styles with element_*()

Now we have some questions…

What are the options for different element functions?

Which element_*() should I use? Check the docs!

How can we modify the legend background, for example?

theme(... = ...)

Theme practice: legend background

legend.background takes an element_rect() and we have the following options available:

args(element_rect)
function (fill = NULL, colour = NULL, linewidth = NULL, linetype = NULL, 
          color = NULL, inherit.blank = FALSE, size = deprecated())

The options are similar to aesthetics in geom layers:

  • fill: Rectangle fill color
  • color/colour: Outline color
  • linetype: Outline line type
  • linewidth: Outline line thickness

Theme practice: legend background

bar_plot +
  theme(legend.background = element_rect(fill = "red", color = "blue"))

Styles with specific choices

But not all aspects of the plot are about the appearance of an element:

  • aspect.ratio: The aspect ratio of a plot
  • legend.position: The positioning of plot legend
  • title.position: The positioning of plot title
  • plot.margin: Margins (padding) around the plot

These are exceptions that don’t take element_*() specifications.

  • What does the documentation say about legend.position?

Theme practice: legend position

bar_plot +
  theme(legend.position = "none")

Theme practice: legend position

bar_plot +
  theme(legend.position = "top")

Theme practice: legend position

bar_plot +
  theme(legend.position = "right")

Theme practice: legend position

bar_plot +
  theme(legend.position = "bottom")

Theme practice: legend position

bar_plot +
  theme(legend.position = "left")

Theme practice: legend position

bar_plot +
  theme(legend.position = c(.5, .5))

Theme practice: legend position

bar_plot +
  theme(legend.position = c(.2, .8))

Interim summary

What we’ve seen so far:

  • You specify the style of specific components of your plot using theme()
  • Themes are hierarchical, and nesting is shown by the "." in the argument name
  • For the most part, styles are specified using element_*() functions
  • Always consult the documentation! https://ggplot2.tidyverse.org/reference/theme.html

Last few minutes

I want to quickly showcase some:

  • Complete themes

  • Theme extensions

  • Additional resources

Complete themes

bar_plot

Complete themes

bar_plot +
  theme_minimal()

Complete themes

bar_plot +
  theme_classic()

Complete themes

bar_plot +
  theme_grey() # Default

Complete themes

Additional feature: scaling with base_size

bar_plot +
  theme_classic(base_size = 5)

Complete themes

Additional feature: scaling with base_size

bar_plot +
  theme_classic(base_size = 15)

Theme extensions

Theme extensions (mostly) come in one of two forms:

  • Custom complete themes

  • Custom theme element_*()s

You can search for ggplot2 theme extensions like you would any other ggplot2 extension package:

Complete theme extensions

Complete theme extensions

library(ggthemes)
bar_plot +
  theme_wsj()

Modify on top of complete themes

library(ggthemes)
bar_plot +
  theme_wsj(base_size = 10) +
  theme(legend.title = element_blank()) # can substitute any element

Theme element_*() extensions

library(elementalist) # devtools::install_github("teunbrand/elementalist")
bar_plot +
  theme(axis.line.x = element_line_wiggle(color = "red", amount = 5),
        legend.background = element_rect_wiggle(color = "black"))

Theme element_*() extensions

library(elementalist) # devtools::install_github("teunbrand/elementalist")
bar_plot +
  theme(axis.line.x = element_line_wiggle(color = "red", amount = 5),
        legend.background = element_rect_wiggle(color = "black"))

Resources

Some further resources:

yjunechoe.github.io/ggplot2-theme-intro