How do I add a variable to the legend without including it in the graph?

Question

I would like to add a value to the legend without adding it to the plot.

library(tidyverse)

gear_color <- c('black', "blue", "orange")
names(gear_color) <- c('3', '4', '5')

mtcars %>% 
  mutate(gear = factor(gear),
         vs = factor(vs)) %>% 
  filter(gear != 4) %>% 
  ggplot(aes(vs, fill = gear)) +
  geom_bar() +
  facet_wrap(~gear) +
  scale_fill_manual(values = gear_color)

How would I add Gear 4 to the legend without adding it to the plot?

Answer 1

The scale_fill_manual can take drop as parameter ( ?scale_fill_manual )

The ... - arguments passed onto discrete_scale

drop - Should unused factor levels be omitted from the scale? The default, TRUE, uses the levels that appear in the data; FALSE uses all the levels in the factor.

So, we can add drop = FALSE in the last line

... +
    scale_fill_manual(values = gear_color, drop = FALSE)

As @eipi10 mentioned in the comments, an empty facet for gear = 4 can be added similarly by include drop = FALSE in facet_wrap

-output