CODEWITHSUNDEEP
rdplyr
ℕʘʘḆḽḘ
ℕʘʘḆḽḘ

Reputation: 19375

R: How to extract a list from a dataframe?

Consider this simple example

> weird_df <- data_frame(col1 =c('hello', 'world', 'again'),
+                       col_weird = list(list(12,23), list(23,24), NA))
> 
> weird_df
# A tibble: 3 x 2
   col1  col_weird
  <chr>     <list>
1 hello <list [2]>
2 world <list [2]>
3 again  <lgl [1]>

I need to extract the values in the col_weird. How can I do that? I see how to do that in Python but not in R. Expected output is: 

> good_df
# A tibble: 3 x 3
   col1   tic   toc
  <chr> <dbl> <dbl>
1 hello    12    23
2 world    23    24
3 again    NA    NA

Upvotes: 6

Views: 8999

Answers (5)

aosmith
aosmith

Reputation: 36076

If you collapse the list column into a string you can use separate from tidyr. I used map from purrr to loop through the list column and create a string with toString.

library(tidyr)
library(purrr)

weird_df %>%
     mutate(col_weird = map(col_weird, toString ) ) %>%
     separate(col_weird, into = c("tic", "toc"), convert = TRUE)

# A tibble: 3 x 3
   col1   tic   toc
* <chr> <int> <int>
1 hello    12    23
2 world    23    24
3 again    NA    NA

You can actually use separate directly without the toString part but you end up with "list" as one of the values.

weird_df %>%
     separate(col_weird, into = c("list", "tic", "toc"), convert = TRUE) %>%
     select(-list)

This led me to tidyr::extract, which works fine with the right regular expression. If your list column was more complicated, though, writing out the regular expression might be a pain.

weird_df %>%
     extract(col_weird, into = c("tic", "toc"), regex = "([[:digit:]]+), ([[:digit:]]+)", convert = TRUE)

Upvotes: 4

ℕʘʘḆḽḘ
ℕʘʘḆḽḘ

Reputation: 19375

well, I came up with a simple one

> weird_df %>% 
+   rowwise() %>%
+   mutate(tic = col_weird[[1]],
+          tac = ifelse(length(col_weird) == 2, col_weird[[2]], NA)) %>% 
+   select(-col_weird) %>% ungroup()
# A tibble: 3 x 3
   col1   tic   tac
  <chr> <dbl> <dbl>
1 hello    12    23
2 world    23    24
3 again    NA    NA

Upvotes: 2

akrun
akrun

Reputation: 886938

Here is one option to do with purrr/tidyverse/reshape2. We unlist the 'col_weird' within map to get the output as list, set the names of the list with 'col1', melt to 'long' format, grouped by 'L1', create a 'rn' column and spread it back to 'wide'

library(tidyverse)
library(reshape2)
weird_df$col_weird %>%
     map(unlist) %>% 
     setNames(., weird_df$col1) %>%
     melt %>% 
     group_by(L1) %>%
     mutate(rn = c('tic', 'toc')[row_number()]) %>%
     spread(rn, value) %>%
     left_join(weird_df[-2], ., by = c(col1 = "L1"))

Upvotes: 2

Enrique P&#233;rez Herrero
Enrique P&#233;rez Herrero

Reputation: 3882

weird_df <- data_frame(col1 = c('hello', 'world'),
                   col_weird = list(list(12,23), list(23,24)))

library(dplyr)
weird_df %>%
  dplyr::mutate(tic = unlist(magrittr::extract2(col_weird, 1)),
                toc = unlist(magrittr::extract2(col_weird, 2)),
                col_weird = NULL)

With the last changes: Note that now col_weird contains list(NA, NA)

weird_df <- data_frame(col1 = c('hello', 'world', 'again'),
                  col_weird = list(list(12,23), list(23,24), list(NA, NA)))

library(dplyr)
weird_df %>%
 dplyr::mutate(col_weird = matrix(col_weird),
 tic = sapply(col_weird, function(x) magrittr::extract2(x, 1)),
 toc = sapply(col_weird, function(x) magrittr::extract2(x, 2)),
 col_weird = NULL)

Upvotes: 2

psychOle
psychOle

Reputation: 1064

You can do this with basic R, thanks to I():

weird_df <- data.frame(col1 =c('hello', 'world'), 
   col_weird = I(list(list(12,23),list(23,24))))

weird_df
>    col1 col_weird
  1 hello    12, 23
  2 world    23, 24

Upvotes: 2

Related Questions