Ugly figure competition

Background: theme()

Every element of a ggplot’s non-data appearance — text, gridlines, backgrounds, legends, axis ticks — is controlled by the theme. You can add complete themes like theme_minimal() or theme_bw() to change the overall look of a plot in one line. The theme() function lets you go further and control individual elements of that look yourself.

theme() defines a collection of parameter, one for each element you want to customize. Many parameters are set using an element_*() function that matches the type of thing you’re changing:

  • element_text() for text (axis labels, titles, legend text…)
  • element_line() for lines (axis lines, gridlines…)
  • element_rect() for rectangles (plot background, panel background, legend background…)
  • element_blank() to remove an element entirely

For example:

ggplot(data = mtcars, aes(x = wt, y = mpg)) +
  geom_point() +
  theme(
    panel.background = element_rect(fill = "cornsilk"),
    axis.title = element_text(color = "darkgreen", face = "bold"),
    axis.text = element_text(size = 14, angle = 45),
    panel.grid.major = element_line(color = "orange"),
    panel.grid.minor = element_blank()
  )

A few things worth noticing:

  • theme() is additive — it changes only the elements you name, leaving everything else at its default (or at whatever a complete theme like theme_minimal() set previously).
  • Order matters. If you call a complete theme (e.g. theme_bw()) after a theme() call, it will reset everything theme() changed. Complete themes should generally come first, with theme() tweaks layered on afterward.
  • There are dozens of themeable elements, covering nearly everything on the plot that isn’t data. The ggplot2 theme vignette lists them all.

Challenge: make it hideous

Create a script for this challenge called day7am-ugly-figure.R. Copy-paste the code from the day 6 end-of-day practice you wrote to make the boxplot in Exercise 1 — percent cover by habitat and species.

Your task: using only theme(), turn this figure into the gnarliest, most visually upsetting version of itself that you can. The underlying geom_boxplot() call and data should stay the same.

Note

Some theme() elements worth abusing:

  • panel.background, plot.background, legend.background — fill colors
  • panel.grid.major, panel.grid.minor — gridline color, thickness, or removal
  • axis.text, axis.title, plot.title, legend.text — font color, size, angle, face
  • legend.position — try somewhere it doesn’t belong
  • axis.ticks — color and length

Rules

  1. The plot must still run without errors and must still technically display the correct data.
  2. All changes must be made through theme() — no changing geom_boxplot() to something else, no fabricating data.
  3. Anything goes otherwise. Clashing colors, unreadable text, chaotic gridlines — go big.

Submission

Find your slide in this Slides presentation. Put your code and your ugly figure on it.