How to customize the plot.background in ggplot with multiple colors

Question

I want to customize the plot background into different colors. Specifically, I want the background to be navy at the top and gray everywhere else, even though I want the plot itself to be white (i.e. overlaid onto the background).

I've tried making the background a custom image, but couldn't get that to work either. Here's the closest visual approximation I've managed:

library(ggplot2)
library(tidyr)
library(grid)

mtcars %>%  
  ggplot(aes(x = wt, y = mpg)) +
  geom_point(color = "red") +
  theme(panel.background = element_rect(fill = "white"),
        plot.background = element_rect(fill = "gray"),
        panel.grid.major = element_line(color = "gray"),
        panel.grid.minor = element_blank(),
        plot.margin = margin(t = 20, r = 10, b = 10, l = 10),
        text = element_text(color = "white")
        )

grid.rect(x = 0.01, y = 1, width = 2, height = 0.2, gp = gpar(fill = "navy", alpha = 1))

But instead of a navy rectangle superimposed on the whole plot, I want it only in the background. Any advice?

Package {cowplot} could be worth a try: https://wilkelab.org/cowplot/articles/drawing_with_on_plots.html . Don't turn to the dark side though ;-) https://en.wikipedia.org/wiki/Chartjunk — I_O, Jul 06 '23 at 15:16

Allan Cameron · Accepted Answer · 2023-07-06T15:57:51.620

In the newer versions of R, we can use linear gradient fills in the grid graphics system on which ggplot2 is built. For example, we can create a gradient fill from gray to navy as follows:

mtcars %>%  
  ggplot(aes(x = wt, y = mpg)) +
  geom_point(color = "red") +
  theme(panel.background = element_rect(fill = "white"),
        plot.background = element_rect(fill = grid::linearGradient(
          c('gray', 'gray', 'navy', 'navy'), stops = c(0, 0.05, 0.95, 1),
        x2 = unit(0, 'npc'))),
        panel.grid.major = element_line(color = "gray"),
        panel.grid.minor = element_blank(),
        plot.margin = margin(t = 20, r = 10, b = 10, l = 10),
        text = element_text(color = "white")
  )

If we don't like the fade and just want a solid bar of navy at the top, we just put the stops of the linear gradient arbitrarily close together at, say, 95% of the height of the plotting window.

mtcars %>%  
  ggplot(aes(x = wt, y = mpg)) +
  geom_point(color = "red") +
  theme(panel.background = element_rect(fill = "white"),
        plot.background = element_rect(fill = grid::linearGradient(
          c('gray', 'gray', 'navy', 'navy'), stops = c(0, 0.949, 0.95, 1),
        x2 = unit(0, 'npc'))),
        panel.grid.major = element_line(color = "gray"),
        panel.grid.minor = element_blank(),
        plot.margin = margin(t = 20, r = 10, b = 10, l = 10),
        text = element_text(color = "white")
  )

As already pointed out in the comments, it is a good idea to avoid unnecessary chart junk like this unless it really helps you to convey the message within the data.

Thanks for your advice keeping charts simple. I'll probably end up scrapping these charts anyway; I just wanted to play around with style. However, I copied each of your code examples and was just left with a white background for both. Is there something I'm missing? — obsequious_octahedron, Jul 06 '23 at 16:24
You need to make sure you are on the latest version of R / grid / ggplot2, and have a suitable graphics device. I use ragg, which you can install with `install.packages('ragg')`. Then use the RStudio menu bar to select Tools -> Global Options -> General -> Graphics -> Backend -> AGG and restart. — Allan Cameron, Jul 06 '23 at 16:27

How to customize the plot.background in ggplot with multiple colors

1 Answers1