mutate_at有两组变量

Question

I just asked a question about generating multiple columns at once with dplyr , and I'm a bonehead and oversimplified the problem and have another question. I'd like to find a dplyr method for dynamically generating columns based on other columns.

  cols <- c("x", "y")
  foo <- c("a", "b")
  bar <- c("c", "d")
  df <- data.frame(a = 1, b = 2, c = 10, d = 20)
  df[cols] <- df[foo] * df[bar]

In my first iteration of the question, I included only one set of previously defined columns, so the following worked:

df %>%
 mutate_at(vars(foo), list(new = ~ . * 5)) %>% 
 rename_at(vars(matches('new')), ~ c('x', 'y')) 

However, as the first few lines of code suggest, I would like to instead multiply two existing columns together, and am unable to figure out how to do this. I have tried:

df %>%
  mutate_at(c(vars(foo), vars(bar)), 
            function(x,y) {x * y})

which returns the error:

Error in (function (x, y)  : argument "y" is missing, with no default

Is it possible to reference multiple sets of columns to be used on each other with mutate_at ?

Answer 1

Well as you want to work with two columns, I think purrr::map2 is the function to work with:

library(purrr)
library(dplyr)

map2(foo, bar, ~ df[[.x]] * df[[.y]]) %>% 
  set_names(cols) %>% 
  bind_cols(df, .)

#>   a b  c  d  x  y
#> 1 1 2 10 20 10 40