changing ggplot2::facet_wrap title from the default

Question

Is there any possible way to change the labels for the facet_wrap variable, as displayed below. So, for example, instead of cyl: 4, cyl: 6, cyl: 8, I want it to read condition: 4, condition: 6, condition: 8. Of course, I can just do this by renaming the variable, but that's not what I want. This is a much simpler version of a custom function where I can't just rename a variable anyway I like.

Another way to put this is, do I have any freedom to label the facet_wrap anyway I like? Kinda like how x aesthetic variable in ggplot2 can have some name (e.g. cyl) in the dataframe (mtcars), but I can still replace it with my own name using labs(x = "cylinder")). I want to have something similar for facet_wrap.

library(dplyr)
library(datasets)
library(ggplot2)
data(mtcars)

# creating a dataframe
df <- dplyr::group_by(mtcars, .dots = c('cyl', 'am')) %>%
  dplyr::summarize(counts = n()) %>%
  dplyr::mutate(perc = (counts / sum(counts)) * 100) %>%
  dplyr::arrange(desc(perc))

# preparing the plot
ggplot2::ggplot(df, aes('', counts)) +
  geom_col(
    position = 'fill',
    color = 'black',
    width = 1,
    aes(fill = factor(am))
  ) +
  facet_wrap(~cyl, labeller = "label_both") + # faceting by `cyl` variable
  geom_label(
    aes(label = paste0(round(perc), "%"), group = factor(am)),
    position = position_fill(vjust = 0.5),
    color = 'black',
    size = 5,
    show.legend = FALSE
  ) +
  coord_polar(theta = "y")

Created on 2018-02-19 by the reprex package (v0.2.0).

Solution 1

To change the facet labels you can provide a named vector of labels to the labeller argument in facet_wrap:

labeller = labeller(cyl = 
    c("4" = "condition: 4",
      "6" = "condition: 6",
      "8" = "condition: 8"))

Here is the full plot code:

ggplot2::ggplot(df, aes('', counts)) +
  geom_col(
    position = 'fill',
    color = 'black',
    width = 1,
    aes(fill = factor(am))
  ) +
  facet_wrap(~cyl, labeller = labeller(cyl = 
    c("4" = "condition: 4",
      "6" = "condition: 6",
      "8" = "condition: 8")
  )) 
  geom_label(
    aes(label = paste0(round(perc), "%"), group = factor(am)),
    position = position_fill(vjust = 0.5),
    color = 'black',
    size = 5,
    show.legend = FALSE
  ) +
  coord_polar(theta = "y")

EDIT based on the comments where a function to return labels is requested:

perhaps something like this:

label_facet <- function(original_var, custom_name){
  lev <- levels(as.factor(original_var))
  lab <- paste0(custom_name, ": ", lev)
  names(lab) <- lev
  return(lab)  
}


ggplot2::ggplot(df, aes('', counts)) +
  geom_col(
    position = 'fill',
    color = 'black',
    width = 1,
    aes(fill = factor(am))
  ) +
  facet_wrap(~cyl, labeller = labeller(cyl = label_facet(df$cyl, "grouping"))) +
geom_label(
  aes(label = paste0(round(perc), "%"), group = factor(am)),
  position = position_fill(vjust = 0.5),
  color = 'black',
  size = 5,
  show.legend = FALSE
) +
  coord_polar(theta = "y")

sessionInfo()

R version 3.4.2 (2017-09-28)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252    LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                           LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] bindrcpp_0.2  ggplot2_2.2.1 dplyr_0.7.4   RMOA_1.0      rJava_0.9-9   RMOAjars_1.0 

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.14      bindr_0.1         magrittr_1.5      munsell_0.4.3     colorspace_1.3-2  R6_2.2.2          rlang_0.1.4      
 [8] plyr_1.8.4        tools_3.4.2       grid_3.4.2        gtable_0.2.0      yaml_2.1.14       lazyeval_0.2.1    assertthat_0.2.0 
[15] digest_0.6.13     tibble_1.4.1      glue_1.2.0        labeling_0.3      compiler_3.4.2    pillar_1.0.1      scales_0.5.0.9000
[22] pkgconfig_2.0.1  

Solution 2

You can do this:

  ggplot2::ggplot(df, aes('', counts)) +
  geom_col(
    position = 'fill',
    color = 'black',
    width = 1,
    aes(fill = factor(am))
  ) +
  facet_wrap(~cyl, labeller = as_labeller(c(`4` = "Condition: 4", `6` = "Condition: 6", `8` = "Condition: 8"))) + # faceting by `cyl` variable
  geom_label(
    aes(label = paste0(round(perc), "%"), group = factor(am)),
    position = position_fill(vjust = 0.5),
    color = 'black',
    size = 5,
    show.legend = FALSE
  ) +
  coord_polar(theta = "y")

Comments

[Indrajeet Patil]

Thanks. This will work for this particular example, but I was looking for a more general solution. Since this is part of a custom function, I don't know a priori how many levels the variable will have (here it's just 4, 6, 8, but there can be more or less no. of levels for another variable) and what kind of name the user will like to give to the facet_wrap variable (here I chose condition, but someone else might want grouping, etc.).

[Indrajeet Patil]

So, basically, I am looking for a function that will create a pie chart from two factor variables. Here is the actual function: github.com/IndrajeetPatil/ggstatsplot/blob/master/R/… It works fine. It's just that, in addition to specifying caption, title, and legend title, I also want the user to have the ability to rename facet_wrap variable anyway they like. Since, as things stand right now, it's always converted to the name condition.

[Indrajeet Patil]

Thanks for the answer though. I will try to think of a way to get its generalized version incorporated into my code.

[Indrajeet Patil]

Only one line in your code needs to be changed and it works! lab <- paste0(custom_name, ": ", lev) Can you please edit the code for future reference or for other readers?

[Indrajeet Patil]

Just tried the modified version with non-numeric factor levels and it worked fine. So, at least based on the initial tests I've run, this seems to be robust.

[Indrajeet Patil]

It's strange that the label positions in the second plot you have provided in your edited answer are all out of whack. Especially look at grouping: 6; those are just wrong labels. That's not happening with me. Are you using a developmental version of ggplot2?

[missuse]

@Indrajeet Patil good catch, I rerun the code on my home PC and the problem is not present, attached my sessionInfo(). Will update with sessionInfo() from work PC. The edited part of my answer was done on it previously.