Given a nested data frame like for example
library(tidyr)
library(dplyr)
m <- mtcars %>% nest_by(cyl)
I want to extract one of the columns in the nested data frames to the top level, creating
more rows but not as many as if I would use m %>% unnest(data)
What works for me and gives me the output I want is
m %>% rowwise(-data) %>% summarise(nest_by(data, gear), .groups = "drop")
However, I feel as if I chose a convoluted road to my target because what I want to achieve is partial un-nesting, and I do it by first making an additional nesting that is then unnested again using the summary(<tibble>) semantics.
I tried m %>% hoist(data, "cyl") but this only got me an additional list column.
Is there a canonical function in dplyr that conveys the intention "partial unnesting" better than my solution?
You don't need rowwise, because nest_by() already gives a rowwise grouped tibble. Since dplyr 1.1.0, summarise should be replaced with reframe which is more appropriate for those situations where multiple rows are returned by group:
m %>% reframe(nest_by(data, gear))
# # A tibble: 8 × 3
# cyl gear data
# <dbl> <dbl> <list<tibble[,9]>>
# 1 4 3 [1 × 9]
# 2 4 4 [8 × 9]
# 3 4 5 [2 × 9]
# 4 6 3 [2 × 9]
# 5 6 4 [4 × 9]
# 6 6 5 [1 × 9]
# 7 8 3 [12 × 9]
# 8 8 5 [2 × 9]