CODEWITHSUNDEEP
rdplyrr-factorforcats
maia-sh
maia-sh

Reputation: 641

Compare factors with different levels, in a dataframe

I am trying to compare two factors within a dataframe to create a new variable. The factors have different levels, which is throwing an error.

Here is a reproducible example

library(dplyr)
library(forcats)

mtcars %>% 
  select(gear, carb) %>% 
  mutate_at(c("gear", "carb"), ~as_factor(.)) %>%
  mutate(gear_vs_carb = gear == carb)

And here is the error:

Error in Ops.factor(gear, carb) : level sets of factors are different

I understand that I can make the comparison by converting the factors to characters or numeric and/or by adding unused levels to the factors to make the levels match, e.g. How can I compare two factors with different levels?

But is it possible to make the comparison directly with the original factors?

The output should look the same as for

mtcars %>% 
  select(gear, carb) %>% 
  mutate(gear_vs_carb = gear == carb)

Thank you for your help!

Upvotes: 1

Views: 848

Answers (2)

G. Grothendieck
G. Grothendieck

Reputation: 269431

You only need to convert one factor to character, not both.

mtcars %>% 
  select(gear, carb) %>% 
  mutate_at(c("gear", "carb"), as_factor) %>%
  mutate(gear_vs_carb = gear == as.character(carb))

Upvotes: 1

akrun
akrun

Reputation: 886938

The == wouldn't work with factor class. One option may be to convert to character and do an elementwise comparison or if the intention is to compare the levels, sort the levels, do the comparison and wrap with all

library(dplyr)
mtcars %>% 
   select(gear, carb) %>% 
   mutate_at(c("gear", "carb"), ~as_factor(.)) %>%
   mutate(gear_vs_carb =  all(sort(levels(gear)) == sort(levels(carb))))
   #or use intersect
   #  mutate(gear_vs_carb = length(intersect(levels(gear), 
   #          levels(carb))) == nlevels(gear))

If we are doing elementwise comparison, convert to character class with as.character and then do the comparison

mtcars %>% 
   select(gear, carb) %>% 
   mutate_at(c("gear", "carb"), ~as_factor(.)) %>%
   mutate(gear_vs_carb = as.character(gear) == as.character(carb))

Upvotes: 1

Related Questions