How to modify the width of a facet_wrap strip?

Question

I have two charts: FigA and FigB. Both are facet wrapped. FigA is facetted based on a short factor label, whereas FigB is faceted based on a much longer factor label. Both FigA and FigB share an x-axis, and so I want to display them vertically (via cowplot) with the strip labels positioned to the right of the plot and with horizontal text.

This results in a lot of white space around FigA's short strip labels.

What is the best way to format the strip label backgrounds so that they expand to fill the available horizontal space?

I've been playing around with strip.background and strip.text margins, size and other parameters, but so far have not produced the desired result. Am I missing something obvious?

Below is a minimum example

(I realise I could probably pivot_longer and produce a single chart with this simple example, but is there a way to modify the strip label directly for more complex situations?):

library(tidyverse)
library(cowplot)
df <- data.frame(   shortCat = sample(c('a','b'), 10, replace=TRUE),
                    longCat = sample(c('a really long label','another really long label'), 10, replace=TRUE),
                    x = sample(seq(as.Date('2020/01/01'), as.Date('2020/12/31'), by="day"), 10),
                    y = sample(0:25, 10, replace = TRUE) )

figA <- df %>% ggplot( aes(x=x,y=y) ) +      
    geom_line() + 
    facet_wrap(vars(shortCat), ncol=1, strip.position ="right", scales="free_y") + 
    theme(  axis.title.y=element_blank(),
            axis.title.x=element_blank(),
            axis.text.x=element_blank(),
            axis.ticks.x=element_blank(), 
            strip.text.y.right = element_text(angle = 0, hjust=0) )

figB <- df %>% ggplot( aes(x=x,y=y) ) +      
    geom_bar(stat="identity") + 
    facet_wrap(vars(longCat), ncol=1, strip.position ="right", scales="free_y") + 
    theme(  axis.title.y=element_blank(),
            strip.text.y.right = element_text(angle = 0, hjust=0) )

plot_grid(figA,figB, ncol=1, align="v")

user_496488 · Answer

For cases where you need to stack two separate plots, here's a way to match the strip widths:

Load the ggtext package, which allows us to have multiple text colors within a single strip label (using html tags).
Get the full text of the longest strip label. Add that as a second line to the short strip labels (we'll add an html tag to do this). This text will ensure that the strip width is exactly the same in both plots.
However, we don't want the extra line of text to be visible. So we'll add some html tags to set the color of this text to be the same as the background fill color of the strip.

library(tidyverse)
library(patchwork) # For plot layout
library(ggtext) # For multiple text colors in strip labels
theme_set(theme_bw())

set.seed(2)
df <- data.frame(shortCat = sample(c('a','b'), 10, replace=TRUE),
                 longCat = sample(c('a really long label','another really long label'), 10, replace=TRUE),
                 x = sample(seq(as.Date('2020/01/01'), as.Date('2020/12/31'), by="day"), 10),
                 y = sample(0:25, 10, replace = TRUE) )

# Test method's robustness by making labels of different lengths
df = df %>% 
  mutate(shortCat2 = gsub("b", "medium label", shortCat))

# Get text of longest label
pad = df$longCat[which.max(nchar(df$longCat))]

# Get colour of strip background
txt.col = theme_get()$strip.background$fill

# Set padding text to same colour as background, so it will be invisible
#  (you can set the color to "white" for a visual confirmation of what this does)
df$shortCat2 = paste0(df$shortCat2,  "
", pad ,"")

figA = df %>%
  ggplot( aes(x=x,y=y) ) +
  geom_line() +
  facet_wrap(vars(shortCat2), ncol=1, strip.position ="right", scales="free_y") +
  theme(axis.title.y=element_blank(),
        axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank(),
        strip.text.y=element_textbox())

figA / figB

Here's another approach in which we add whitespace to expand the strip labels to the desired width. It requires a monospace font, which makes it less flexible, but it doesn't require the use of html tags as in the previous method.

# Identify monospaced fonts on your system
fonts = systemfonts::system_fonts()
fonts %>% filter(monospace) %>% pull(name)
#>  [1] "Menlo-Bold"                "Courier-Oblique"          
#>  [3] "Courier-BoldOblique"       "AppleBraille"             
#>  [5] "AppleBraille-Pinpoint8Dot" "AndaleMono"               
#>  [7] "Menlo-BoldItalic"          "Menlo-Regular"            
#>  [9] "CourierNewPS-BoldMT"       "AppleBraille-Outline6Dot" 
#> [11] "GB18030Bitmap"             "Monaco"                   
#> [13] "AppleBraille-Outline8Dot"  "PTMono-Regular"           
#> [15] "PTMono-Bold"               "AppleColorEmoji"          
#> [17] "Menlo-Italic"              "CourierNewPS-ItalicMT"    
#> [19] "Courier"                   "Courier-Bold"             
#> [21] "CourierNewPSMT"            "AppleBraille-Pinpoint6Dot"
#> [23] "CourierNewPS-BoldItalicMT"

# Set theme to use a monospace font
theme_set(theme_bw() + 
            theme(text=element_text(family="Menlo-Regular")))

figA <- df %>%
  mutate(shortCat = paste0(shortCat, 
                           paste(rep(" ", max(nchar(longCat)) - 1), collapse=""))
  ) %>%
  ggplot( aes(x=x,y=y) ) +
  geom_line() +
  facet_wrap(vars(shortCat), ncol=1, strip.position ="right", scales="free_y") +
  theme(axis.title.y=element_blank(),
        axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank(),
        strip.text.y.right = element_text(angle = 0, hjust=0))

figB <- df %>% ggplot( aes(x=x,y=y) ) +
  geom_line() +
  facet_wrap(vars(longCat), ncol=1, strip.position ="right", scales="free_y") +
  theme(  axis.title.y=element_blank(),
          strip.text.y.right = element_text(angle = 0, hjust=0) )

figA / figB

If you can reshape your data to long format and make a single plot, that's the most straightforward approach:

theme_set(theme_bw())

df %>% 
  pivot_longer(matches("Cat")) %>% 
  mutate(value = fct_relevel(value, "a", "b")) %>% 
  ggplot(aes(x,y)) +
  geom_line() +
  facet_wrap(~value, ncol=1, strip.position="right") +
  theme(strip.text.y=element_text(angle=0, hjust=0))

How to modify the width of a facet_wrap strip?

1 Answers