CODEWITHSUNDEEP
rdplyrtidyeval
Matiasko
Matiasko

Reputation: 187

Passing two sets of column names to function

I am trying to pass two sets of column names to function and do something with them using dplyr. Normally for one set I would use the ellipsis (...) and convert it to quosures with enquos(). But now I have two sets of column names so I thought about using lists to store them. How should I make this work in the most efficient way? (Answers using purrr, rlang and any other packages' functions are most welcome)

Packages and example of data

library(dplyr) #I use whole library(tidyverse) but this is sufficient for this case

some.data <- tibble(col1 = sample(letters[1:3], 500, replace = T),
                    col2 = sample(letters[1:3], 500, replace = T),
                    col3 = sample(letters[4:6], 500, replace = T),
                    col4 = sample(letters[4:6], 500, replace = T))

My function (put simple) looks like this:

cross_table <- function(data = NULL, list1 = NULL, list2 = NULL){

   for(l1 in list1){
      for(l2 in list2){

         data.out <- data %>% 
            count(l1, l2) %>% 
            spread(l2, n, fill = 0, drop = FALSE)

         print(data.out) #Just to show it works. I want to use 'data.out' object later on

      }
   }
}

and I want to use function like this (not giving columns' names as strings)

some.data %>%
   cross_table(list1 = list(col1, col2), list2 = list(col3, col4))

Upvotes: 2

Views: 81

Answers (1)

aosmith
aosmith

Reputation: 36076

The vars() function could be a good fit here. You'd use it in place of list() in your function arguments. I saw an example in this SO answer and it extends to your situation pretty handily.

That plus some tidyeval within your loop would look like:

cross_table <- function(data = NULL, list1 = NULL, list2 = NULL){

        for(l1 in list1){
            for(l2 in list2){

                l1 = enquo(l1)
                l2 = enquo(l2)
                data.out <- data %>%
                    count(!!l1, !!l2) %>%
                    spread(!!l2, n, fill = 0, drop = FALSE) 

                print(data.out) 
            }
        }
    }

some.data %>%
   cross_table(list1 = vars(col1, col2), list2 = vars(col3, col4))


# A tibble: 3 x 4
  col1      d     e     f
  <chr> <dbl> <dbl> <dbl>
1 a        58    61    53
2 b        38    59    47
3 c        65    59    60
# A tibble: 3 x 4
  col1      d     e     f
  <chr> <dbl> <dbl> <dbl>
1 a        53    61    58
2 b        44    47    53
3 c        56    62    66
# A tibble: 3 x 4
  col2      d     e     f
  <chr> <dbl> <dbl> <dbl>
1 a        55    60    51
2 b        57    67    56
3 c        49    52    53
# A tibble: 3 x 4
  col2      d     e     f
  <chr> <dbl> <dbl> <dbl>
1 a        51    56    59
2 b        63    55    62
3 c        39    59    56

You can also use alist() in place of list() (which it looks like I learned at one point but have since forgotten :-D).

Upvotes: 3

Related Questions