Integrate fct_other functionality into fct_collapse

[jmpsteen]

Thanks for this great package! I'm wondering though whether there would be any possibility of integrating features of fct_other into the fct_collapse function?

I currently need to recode a factor by collapsing a targeted set of levels (say c("A", "B", "C", "D", "E")) into new levels (say c("AB", "CDE")) and I want to recode all non-targeted levels (which usually is quite a long set of levels to manually specify, e.g. all other letters of the alphabet) into an "other" category. In other words: I'ld like fct_collapse to have an additional argument (similar to fct_other's other_level argument) that, when specified, automatically identifies all non-targeted levels (i.e. those not specified in any of the vectors, which in fct_other would be specified in the drop argument) and collapses these into a category with level name specified in this argument.

This feature would be quite helpful, because the only solution I can think of now is quite cumbersome. That is, I have to manually specify all new (collapsed) factor level names (entered in the fct_collapse call) in the keep argument of fct_other, while I assume these could easily be internally identified automatically in the fct_collapse function.

[dholstius]

+1 for this. I often have use for a tabulation where the final element is “(other)”.

Lexically, the leading parenthesis makes “(other)” follow alphanum levels, so I’d suggest that as a default.

[hadley]

Can you please provide a minimal reprex (reproducible example)? The goal of a reprex is to make it as easy as possible for me to recreate your problem so that I can fix it: please help me help you!

If you've never heard of a reprex before, start by reading "What is a reprex", and follow the advice further down the page. Please make sure your reprex is created with the reprex package as it gives nicely formatted output and avoids a number of common pitfalls.

[BirgerNi]

+1 for this.

library(forcats)

set.seed(123)
data <- sample(LETTERS[1:10], 99, replace = TRUE) 


# What I had to do
data %>% 
  fct_collapse(AB = c("A", "B"), CDE = c("C", "D", "E")) %>%
  fct_other(keep = c("AB", "CDE"))
#>  [1] CDE   Other CDE   Other Other AB    Other Other Other CDE   Other
#> [12] CDE   Other Other AB    Other CDE   AB    CDE   Other Other Other
#> [23] Other Other Other Other Other Other CDE   AB    Other Other Other
#> [34] Other AB    CDE   Other CDE   CDE   CDE   AB    CDE   CDE   CDE  
#> [45] AB    AB    CDE   CDE   CDE   Other AB    CDE   Other AB    Other
#> [56] CDE   AB    Other Other CDE   Other AB    CDE   CDE   Other CDE  
#> [67] Other Other Other CDE   Other Other Other AB    CDE   CDE   CDE  
#> [78] Other CDE   AB    CDE   Other CDE   Other AB    CDE   Other Other
#> [89] Other AB    AB    Other CDE   Other CDE   AB    Other AB    CDE  
#> Levels: AB CDE Other


# What I like to do
# data %>%
#   fct_collapse(AB = c("A", "B"), CDE = c("C", "D", "E"), group_other = TRUE)

[batpigandme]

@BirgerNi thanks for the reprex.
FYI, re +1 (for @holstius as well), see our new article re. upvoting issues/FRs on github 👍:
https://www.tidyverse.org/articles/2018/04/upvoting-issues/

[hadley]

Seems like a reasonable feature - the main thing to figure out is the name of the argument

[AmeliaMN]

I'll try this one!