How to use dplyr functions such as filter inside nested data frames with map

Question

I'm want to use map() of purrr package to apply filter() function to the data stored in a nested data frame.

Below is my code, where I would filter and then nest it.

I have achieved my desired result by performing nest() twice i.e. once on all the complete data and the second time on filtered data.

library(tidyverse)

df <- tibble(
  x = sample(x = rep(c('a','b'),5), size = 10),
  y = sample(c(1:10)),
  z = sample(c(91:100))
) 
 #First time nested  
df_fully_nested_dataframe <- df %>% 
  group_by(x) %>% 
  nest(.key = 'full')
#Second time nested
df_filtered_nested_dataframe <- df %>%
  filter(z >= 95) %>%  
  group_by(x) %>% 
  nest(.key = 'filtered')

#The desired outcome is a single data frame with 2 nested list-columns: one full data and other with filtered data.
  ## Now how to achieve this without dividing into 2 dataframes  
df_nested <- df_fully_nested_dataframe %>% 
  left_join(df_filtered_nested_dataframe, by = 'x')

The objects looks as follows:

> df
# A tibble: 10 x 3
       x     y     z
   <chr> <int> <int>
 1     b     8    93
 2     a     9    94
 3     b    10    99
 4     a     5    97
 5     b     2   100
 6     b     3    95
 7     a     7    96
 8     b     6    92
 9     a     4    91
10     a     1    98

> df_fully_nested_dataframe
 # A tibble: 2 x 2 
      x             full
  <chr>           <list>
1     b <tibble [5 x 2]>
2     a <tibble [5 x 2]>

> df_filtered_nested_dataframe
# A tibble: 2 x 2
      x         filtered
  <chr>           <list>
1     b <tibble [3 x 2]>
2     a <tibble [3 x 2]>

> df_nested
# A tibble: 2 x 3
      x             full         filtered
  <chr>           <list>           <list>
1     b <tibble [5 x 2]> <tibble [4 x 2]>
2     a <tibble [5 x 2]> <tibble [4 x 2]>

Answer 1

You can use map() call as follows:

 map(full, ~ filter(., z >= 95))

where . stands for individual nested tibble, to which you can apply the filter directly

df_nested_again <- df_fully_nested_dataframe %>% mutate(filtered = map(full, ~ filter(., z >= 95)))

identical(df_nested, df_nested_again)
# [1] TRUE