Search code examples
rggplot2labelboxplotfacet

ggplot2 - Manually specifying boxplot facet labels as symbol/math expressions

I have a tibble data frame called my_data that looks like this:

> my_data
# A tibble: 60 x 4
   SPECIES       simulation_id psi_hat p_hat
   <chr>                 <int>   <dbl> <dbl>
 1 Grey squirrel            74   0.527 0.306
 2 Grey squirrel           102   0.526 0.316
 3 Grey squirrel           142   0.527 0.309
 4 Grey squirrel           121   0.527 0.309
 5 Grey squirrel            25   0.526 0.317
 6 Grey squirrel            50   0.527 0.309
 7 Grey squirrel            67   0.491 0.326
 8 Grey squirrel            19   0.527 0.306
 9 Grey squirrel           174   0.527 0.302
10 Grey squirrel            46   0.527 0.309
# ... with 50 more rows

There are three different SPECIES in my_data.

I want to make a boxplot of the psi_hat and p_hat variables for each SPECIES. I started by preparing my tibble like so:

my_data_plot <- my_data %>% 
    gather("psi_hat", "p_hat", key = "parameter", value = "estimate")

Then plotting it like this:

ggplot(data = my_data_plot, mapping = aes(x = SPECIES)) +
    geom_boxplot(mapping = aes(y = estimate)) + 
    facet_grid(. ~ parameter)

Which gives me:

enter image description here

What I want to do now is to label the facets as actual Greek symbols rather than e.g. "psi_hat" so that "psi_hat" would get the symbol equivalent to expression(hat(psi)) in the facet label.

I am aware that in facet_grid I could use labeller = label_parse, but that means my variable names have to be actual expressions, but I don't know how to do that when the variable names have to be a function call (i.e. hat(psi)). For example, I tried my_data <- my_data %>% rename(hat(psi) = "psi_hat") which gives me this error:

Error: unexpected '=' in:
"my_data <- my_data %>% 
    rename(hat(psi) ="

I also considered labeller = bquote, but that seems to require me to specify rows and columns when I only have columns in this plot.

So, I am wondering if there is a need, or way, for me to manually define expressions for those facet labels? Or is there a more algorithmic way of achieving this?

Thank you!

Here is the dput() of my_data:

structure(list(SPECIES = c("Grey squirrel", "Grey squirrel", 
"Grey squirrel", "Grey squirrel", "Grey squirrel", "Grey squirrel", 
"Grey squirrel", "Grey squirrel", "Grey squirrel", "Grey squirrel", 
"Grey squirrel", "Grey squirrel", "Grey squirrel", "Grey squirrel", 
"Grey squirrel", "Grey squirrel", "Grey squirrel", "Grey squirrel", 
"Grey squirrel", "Grey squirrel", "Red fox", "Red fox", "Red fox", 
"Red fox", "Red fox", "Red fox", "Red fox", "Red fox", "Red fox", 
"Red fox", "Red fox", "Red fox", "Red fox", "Red fox", "Red fox", 
"Red fox", "Red fox", "Red fox", "Red fox", "Red fox", "Roe Deer", 
"Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", 
"Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", 
"Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", "Roe Deer", 
"Roe Deer"), simulation_id = c(74L, 102L, 142L, 121L, 25L, 50L, 
67L, 19L, 174L, 46L, 162L, 31L, 199L, 147L, 163L, 18L, 170L, 
172L, 106L, 157L, 158L, 200L, 187L, 103L, 144L, 77L, 78L, 139L, 
75L, 57L, 171L, 195L, 90L, 136L, 109L, 30L, 152L, 97L, 74L, 8L, 
161L, 73L, 130L, 151L, 150L, 184L, 198L, 177L, 97L, 88L, 123L, 
190L, 83L, 75L, 143L, 159L, 103L, 16L, 200L, 2L), psi_hat = c(0.526863413764263, 
0.525835491514182, 0.526515540307052, 0.526515540305866, 0.525668605170743, 
0.526515540309525, 0.491205525300549, 0.526863413760793, 0.527216807861263, 
0.526515540305866, 0.526515540306427, 0.526003604543373, 0.526003604547465, 
0.527039406294591, 0.491512271725946, 0.526343604414775, 0.526515540311415, 
0.490421555053814, 0.526003604541449, 0.526688801109048, 0.79766843868635, 
0.801911295111649, 0.797668438685209, 0.79766843868356, 0.736091175845588, 
0.801911294988272, 0.797668438684876, 0.797668438686537, 0.764665429923081, 
0.806326437262252, 0.769630414385322, 0.806326437300003, 0.733141462334417, 
0.801911295128621, 0.777085591061084, 0.801911295171974, 0.703578316906639, 
0.777085591642251, 0.793615265014951, 0.806326437459794, 0.662677377511043, 
0.701654360968483, 0.704629679152389, 0.70622149033349, 0.663754148939585, 
0.704629679094459, 0.705416487515374, 0.677464341430293, 0.707045218951227, 
0.662338556587023, 0.674456696561803, 0.701654360967044, 0.639650521202062, 
0.625380503760432, 0.704629679094605, 0.664561357366103, 0.704629679143759, 
0.636859548660175, 0.66375414894085, 0.707045219029558), p_hat = c(0.305763297214646, 
0.315819064091298, 0.309115591855344, 0.309115591855175, 0.317494717397372, 
0.309115591855145, 0.326303719905328, 0.305763297214592, 0.302410595448716, 
0.309115591855175, 0.309115591855139, 0.314143327866378, 0.314143327866402, 
0.304086998925611, 0.322825888575185, 0.310791594807592, 0.309115591855151, 
0.335654773573103, 0.314143327866529, 0.307439493743849, 0.0690109229482788, 
0.0677020110780009, 0.0690109229483345, 0.0690109229474338, 0.0744703724155286, 
0.0677020110840996, 0.0690109229257095, 0.0690109229256593, 0.0704604468308165, 
0.0663903378649822, 0.0691283355131073, 0.0663903378091044, 0.0737444081467713, 
0.0677020110873479, 0.0665479524846632, 0.0677020110750217, 0.073966658329169, 
0.0665479525245426, 0.0703192747621257, 0.0663903375835434, 0.166196509352831, 
0.157146399287494, 0.152026850762839, 0.14946227707946, 0.163675003941755, 
0.152026850344473, 0.150744974403108, 0.15740696035152, 0.148178744896067, 
0.167783160277052, 0.162569260757996, 0.157146399287514, 0.162501076719809, 
0.177160578693552, 0.152026850344548, 0.162149934290892, 0.152026850679299, 
0.168221557280052, 0.163675003941561, 0.148178745461308)), class = c("tbl_df", 
"tbl", "data.frame"), row.names = c(NA, -60L))
Solution

  • You can try this. Because your data is in long format, I use case_when to rename the variables with an expression to be read by the labeller

    library(tidyverse)

df %>% 
    gather(key = "parameter", value = "estimate", -SPECIES, -simulation_id) %>%
  mutate(parameter = case_when(
    parameter == "psi_hat" ~ "hat(psi)",
    parameter == "p_hat" ~ "hat(p)"
  )) %>%
  ggplot(aes(x = SPECIES)) +
    geom_boxplot(mapping = aes(y = estimate)) + 
    facet_grid(. ~ parameter, labeller = label_parsed)

    Created on 2018-08-18 by the reprex package (v0.2.0).