CODEWITHSUNDEEP
rtidyversepurrrtibble
FMM
FMM

Reputation: 2015

Apply function at different levels of a tibble

I have the following tibble:

library(tidyverse)

set.seed(1234)

df <- tibble(
  x1 = letters[1:2],
  y1 = list(
    tibble(
      x2 = letters[3:4],
      y2 = list(
        tibble(
          x3 = seq(1, 100, 1),
          y3 = rnorm(100)
        )
        )
      )
    )
)

And I need to access the tibble inside the tibble that contains x3 and y3 and apply a custom function to each data frame. For simplicity, let's say I need to apply base::mean() to y3.

My real data is much bigger than this, so I am looking for a clean and efficient way of doing it. Any ideas?

Upvotes: 2

Views: 55

Answers (2)

Nettle
Nettle

Reputation: 3321

Could you just unnest your way down?

 df %>% unnest() %>% unnest() %>% 
   group_by(x2) %>% 
   summarise(mean(y3))

# A tibble: 2 x 2
  x2    `mean(y3)`
  <chr>      <dbl>
1 c         -0.157
2 d         -0.157

Not sure how you want your final dataframe to look, but here's another suggestion

 df %>% unnest() %>% 
   mutate(y3.average = map(y2, ~mean(.$y3)) ) %>% 
   unnest(y3.average)

# A tibble: 4 x 4
  x1    x2    y2                 y3.average
  <chr> <chr> <list>                  <dbl>
1 a     c     <tibble [100 × 2]>     -0.157
2 a     d     <tibble [100 × 2]>     -0.157
3 b     c     <tibble [100 × 2]>     -0.157
4 b     d     <tibble [100 × 2]>     -0.157

Upvotes: 2

Rich Pauloo
Rich Pauloo

Reputation: 8402

You need to locate the level at which you want to apply the function (which I do through bracket indexing), and then apply the function. I hope this is transferable to what you need to do. 

> df[["y1"]][[1]][[2]] %>% lapply(., function(x){mean(x$y3)})

[[1]]
[1] 0.04124318

[[2]]
[1] 0.04124318

Upvotes: 0

Related Questions