How to filter elements of a list column

Question

I have a data frame (df) with 3 variables:

• ID (1st id variable)
• ID2 (2nd id variable)

• data (list column created with the tidyr::nest() function.

  library(tidyverse)
  library(lubridate)
  
  dates <- ymd(c('2018-02-01', '2018-02-06', '2018-02-10', 
                 '2018-02-21', '2018-04-05'))
  df.1 <- tibble(ID = paste0('ID_', rep(1, 5)),
             ID2 = LETTERS[1:5],
             DATE = dates) %>%
             group_by(ID) %>% 
             nest()
  
  
   df.2 <- tibble(ID = paste0('ID_', rep(1, 6)),
             ID2 = LETTERS[1:6])
  
  
   df <- df.1 %>%
   left_join(df.2, by = 'ID')
  

The list column (data) contains 2 variables:

• ID2
• DATE

I would like to keep all elements in the list column (data[[DATE]]) for which data[[ID2]] != df$ID2.

Is there any way to apply a filter function -- maybe of the purrr package?

Thanks very much in advance!

Answer 1

We can use map2. We feed the list-column (data) as .x argument and df$ID2 as .y argument and filter on each .x where .x$ID2 != .y:

library(tidyverse)

output <- df %>%
  mutate(data = data %>% map2(ID2, ~ filter(.x, ID2 != .y))) 

Output:

> output
# A tibble: 6 x 3
  ID    data             ID2  
  <chr> <list>           <chr>
1 ID_1  <tibble [4 x 2]> A    
2 ID_1  <tibble [4 x 2]> B    
3 ID_1  <tibble [4 x 2]> C    
4 ID_1  <tibble [4 x 2]> D    
5 ID_1  <tibble [4 x 2]> E    
6 ID_1  <tibble [5 x 2]> F  

> output %>%
+   pull(data)
[[1]]
# A tibble: 4 x 2
  ID2   DATE      
  <chr> <date>    
1 B     2018-02-06
2 C     2018-02-10
3 D     2018-02-21
4 E     2018-04-05

[[2]]
# A tibble: 4 x 2
  ID2   DATE      
  <chr> <date>    
1 A     2018-02-01
2 C     2018-02-10
3 D     2018-02-21
4 E     2018-04-05

[[3]]
# A tibble: 4 x 2
  ID2   DATE      
  <chr> <date>    
1 A     2018-02-01
2 B     2018-02-06
3 D     2018-02-21
4 E     2018-04-05

[[4]]
# A tibble: 4 x 2
  ID2   DATE      
  <chr> <date>    
1 A     2018-02-01
2 B     2018-02-06
3 C     2018-02-10
4 E     2018-04-05

[[5]]
# A tibble: 4 x 2
  ID2   DATE      
  <chr> <date>    
1 A     2018-02-01
2 B     2018-02-06
3 C     2018-02-10
4 D     2018-02-21

[[6]]
# A tibble: 5 x 2
  ID2   DATE      
  <chr> <date>    
1 A     2018-02-01
2 B     2018-02-06
3 C     2018-02-10
4 D     2018-02-21
5 E     2018-04-05