1

I have a nested list with 3 levels:

m = list(try1 = list(list(court = c("jack", "queen", "king"),
                          suit = list(diamonds = 2, clubs = 5)), 
                     list(court = c("jack", "queen", "king"),
                          suit = list(diamonds = 45, clubs = 67))), 
         try2 = list(list(court = c("jack", "queen", "king"),
                          suit = list(diamonds = 400, clubs = 300)), 
                     list(court = c("jack", "queen", "king"),
                          suit = list(diamonds = 5000, clubs = 6000))))

> str(m)
List of 2
 $ try1:List of 2
  ..$ :List of 2
  .. ..$ court: chr [1:3] "jack" "queen" "king"
  .. ..$ suit :List of 2
  .. .. ..$ diamonds: num 2
  .. .. ..$ clubs   : num 5
  ..$ :List of 2
  .. ..$ court: chr [1:3] "jack" "queen" "king"
  .. ..$ suit :List of 2
  .. .. ..$ diamonds: num 45
  .. .. ..$ clubs   : num 67
 $ try2:List of 2
  ..$ :List of 2
  .. ..$ court: chr [1:3] "jack" "queen" "king"
  .. ..$ suit :List of 2
  .. .. ..$ diamonds: num 400
  .. .. ..$ clubs   : num 300
  ..$ :List of 2
  .. ..$ court: chr [1:3] "jack" "queen" "king"
  .. ..$ suit :List of 2
  .. .. ..$ diamonds: num 5000
  .. .. ..$ clubs   : num 6000

For each sublist in try1 and try2, I need to extract the suit sublist and rbind its elements such that the resulting data frame is in a long format with 4 columns - value (the value of the suit), suit (which identifies which suit the value comes from, i.e. diamonds or clubs), iter (to identify which sublist the suit belongs to, i.e. 1 or 2) and try (try1 or try2).

I could achieve this using a combination of expand.grid() and mapply():

grd = expand.grid(try = names(m), iter = 1:2, suit = c("diamonds", "clubs"))

grd$value = mapply(function(x, y, z) m[[x]][[y]]$suit[[z]], grd[[1]], grd[[2]], grd[[3]])

The result:

> grd
   try iter     suit value
1 try1    1 diamonds     2
2 try2    1 diamonds   400
3 try1    2 diamonds    45
4 try2    2 diamonds  5000
5 try1    1    clubs     5
6 try2    1    clubs   300
7 try1    2    clubs    67
8 try2    2    clubs  6000

However, I was wondering if there was a more general/concise way of reproducing the above result (preferably in base R)? I was thinking about extracting the suit element from each sublist and then using something like stack() recursively on the resulting list:

rapply(m, function(x) setNames(stack(x), names(x)))

But this throws an error, I'm not quite sure why and I don't know what to use in its place.

Uwe Keim
  • 39,551
  • 56
  • 175
  • 291
user51462
  • 1,658
  • 2
  • 13
  • 41

1 Answers1

2

We could use a combination of map with melt

library(purrr)
library(reshape2)
library(dplyr)
map_df(m, ~ .x %>%
                 map(pluck, "suit")  %>% 
                   melt, .id = 'try')

Or with enframe and map

library(tibble)
map_df(m, ~ .x %>% 
              map_df(pluck, "suit") %>% 
                    map_df(~ enframe(.x, name = "iter") %>%
                       unnest, .id = "suit"), .id = 'try'  )
# A tibble: 8 x 4
#  try   suit      iter value
#  <chr> <chr>    <int> <dbl>
#1 try1  diamonds     1     2
#2 try1  diamonds     2    45
#3 try1  clubs        1     5
#4 try1  clubs        2    67
#5 try2  diamonds     1   400
#6 try2  diamonds     2  5000
#7 try2  clubs        1   300
#8 try2  clubs        2  6000
akrun
  • 874,273
  • 37
  • 540
  • 662
  • 1
    That's not at all possible with `rapply`, is it? – jay.sf Feb 17 '19 at 13:28
  • 2
    @jay.sf First, the 'suit' needs to be extracted which is nested inside. That is the reason I am using `map` – akrun Feb 17 '19 at 20:19
  • 1
    Thank you @akrun! I had spent ages trying to figure this out so I'm grateful for your help. I was just wondering why `sublist = m %>% map(pluck, "suit")` does not work on its own? `sublist = modify_depth(m, 2, "suit")` worked but `pluck(m, "suit")` returns NULL. – user51462 Feb 17 '19 at 23:52
  • because `modify_depth` `2` is extracting from the nested list, while `map(pluck` is looking for 'suit' in the first level – akrun Feb 18 '19 at 01:56