CODEWITHSUNDEEP
rdplyr
alfalfa
alfalfa

Reputation: 121

Ungroup some but not all variables in R + dplyr

In dplyr with R, if I've grouped by multiple variables can I then ungroup by a subset of those with ungroup? As far as I can tell, ungroup ignores any arguments provided and always ungroups everything.

Example:

mtcars %>%
  group_by(cyl, vs, am) %>%
  # this removes 'am' but grouping by 'cyl' and 'vs' remain
  summarize(n = n()) %>%  
  ungroup(vs)  # I want this to work but it ungroups everything, not just 'vs'

Is there a way to remove the grouping by vs while keeping the grouping by cyl? I can accomplish this via

mtcars %>%
  group_by(cyl, vs, am) %>%
  summarize(n = n()) %>%  
  ungroup() %>%
  group_by(cyl)

but I feel like there has to be a better way than removing both groupings and re-adding just the cyl grouping.

I'm using the summarize in my examples because that's my particular context, but this behavior seems to be evident even in something as minimal as this:

mtcars %>%
  group_by(cyl, vs) %>%
  ungroup(vs)  # ungroups everything, not just 'vs'

Update: Thanks everyone for the great suggestions and comments. I just tried this and it works with no fuss due to add=FALSE in group_by, and it's as clean as what I was hoping ungroup could do:

mtcars %>%
  group_by(cyl, vs, am) %>%
  summarize(n = n()) %>%  
  group_by(cyl)

Thanks to @camille and @julius-vainora I know that ungroup by itself cannot do this, so I'll have to do grouping rather than ungrouping. @RyanD's suggestions were also very helpful.

Upvotes: 2

Views: 1979

Answers (1)

etrowbridge
etrowbridge

Reputation: 405

Answering this question based on present dplyr documentation: https://dplyr.tidyverse.org/articles/grouping.html

"You can also choose to selectively ungroup by listing the variables you want to remove:"

by_sex_gender %>% 
  ungroup(sex) %>% 
  tally()

Upvotes: 2

Related Questions