[SOLVED] Specify levels when `mutate`-ing dataframe variables to factors

Specify levels when `mutate`-ing dataframe variables to factors

Let's say I have the following tibble dataframe called data:

library(tibble)

data <- tribble(
    ~"ID", ~"some factor", ~"some other factor", 
    1L, "low", "high",
    2L, "very high", "low",
    3L, "very low", "low",
    4L, "high", "very high",
    5L, "very low", "very low"
)

I use the fct() function in forcats to convert my two factor variables accordingly:

library(dplyr)
library(forcats)

data <- data %>%
        mutate(across(starts_with("some"), fct))

Which gives me:

# A tibble: 5 × 3
     ID `some factor` `some other factor`
  <int> <fct>         <fct>              
1     1 low           high               
2     2 very high     low                
3     3 very low      low                
4     4 high          very high          
5     5 very low      very low

However, when I call fct this way it's unclear to me how to specify the levels of this ordinal variable. The order I would like is:

order <- c("very low", "low", "high", "very high")

How should I do this with dplyr's set of functions? The goal is to have ggplot2 visualizations that respect this ordering.

Solution

When you use across() you can pass extra arguments along to the called function through across's ....

data <- data %>%
  mutate(across(starts_with("some"), fct, levels = order))

This is equivalent to

data <- data %>%
  mutate(across(starts_with("some"), function(x) fct(x, levels = order)))

(This is a common paradigm in R, many functions where you are applying a function have a ... argument for arguments that will be passed along to the applied function, see also lapply, sapply, purrr::map, etc.)