CODEWITHSUNDEEP
rdataframetidyversepurrr
andhewoulddarkenthememory
andhewoulddarkenthememory

Reputation: 23

purrr: Split dataframe and run map on several columns

I have a large dataframe and I need to calculate values for combinations of columns based on an id column. I would like to do so within a tidyverse framework and I can get there, but it does not seem to be very elegant and error-prone. Perhaps someone could help me out.

Here is a minimal working example similar to the actual data.

library(tidyverse)

df <- tibble(
  id_combo = c("A_A1", "A_A1", "A_A2", "A_A2", "A_A2"),
  f1 = runif(5),
  f2 = runif(5),
  f3 = runif(5),
  b1 = runif(5),
  b2 = runif(5),
  b3 = runif(5)
)

f1_f2 <- df %>%
  split(.$id_combo) %>%
  map_dbl(., ~var(log(.$f1))+var(log(.$f2)))

f1_f3 <- df %>%
  split(.$id_combo) %>%
  map_dbl(., ~var(log(.$f1))+var(log(.$f2))+var(log(.$f3)))

f1_b2 <- df %>%
    split(.$id_combo) %>%
  map_dbl(., ~var(log(.$f1))+var(log(.$f2))+
              var(log(.$b1))+var(log(.$b2)))

f1_b3 <- df %>%
  split(.$id_combo) %>%
  map_dbl(., ~var(log(.$f1))+var(log(.$f2))+var(log(.$f3))+
              var(log(.$b1))+var(log(.$b2))+var(log(.$b3)))

var_sum_df <- tibble(id_combo = names(f1_f2),f1_f2, f1_f3, f1_b2, f1_b3)

What I hope to achieve is to run the map_dbl function (or a sensible equivalent) after split(.$id_combo), specifying the columns on the fly.

I am sure this would be possible, but my R knowledge is not yet advanced enough to figure out myself.

Upvotes: 0

Views: 786

Answers (1)

Cettt
Cettt

Reputation: 11981

I am not sure if I understood the question correctly, but is this what you are looking for?

library(tidyverse)
df %>% group_by(id_combo) %>% 
  summarise_all(~var(log(.x))) %>% 
  mutate(f1_f2 = f1 + f2,
         f1_f3 = f1_f2 + f3, 
         f1_b2 = f1_f3 + b1 + b2,
         f1_b3 = f1_b2 + b3) %>% 
  select(id_combo, contains("_"))

# A tibble: 2 x 5
  id_combo  f1_f2  f1_f3 f1_b2 f1_b3
  <chr>     <dbl>  <dbl> <dbl> <dbl>
1 A_A1     0.0582 0.0701  1.24  6.89
2 A_A2     2.43   2.57    3.50  3.76

Upvotes: 1

Related Questions