在 dplyr mutate_at 调用中使用多列的函数

Question

I'd like to use dplyr's mutate_at function to apply a function to several columns in a dataframe, where the function inputs the column to which it is directly applied as well as another column in the dataframe.

As a concrete example, I'd look to mutate the following dataframe

# Example input dataframe
df <- data.frame(
    x = c(TRUE, TRUE, FALSE),
    y = c("Hello", "Hola", "Ciao"),
    z = c("World", "ao", "HaOlam")
)

with a mutate_at call that looks similar to this

df %>%
mutate_at(.vars = vars(y, z),
          .funs = ifelse(x, ., NA))

to return a dataframe that looks something like this

# Desired output dataframe
df2 <- data.frame(x = c(TRUE, TRUE, FALSE),
                  y_1 = c("Hello", "Hola", NA),
                  z_1 = c("World", "ao", NA))

The desired mutate_at call would be similar to the following call to mutate :

df %>%
   mutate(y_1 = ifelse(x, y, NA),
          z_1 = ifelse(x, z, NA))

I know that this can be done in base R in several ways, but I would specifically like to accomplish this goal using dplyr's mutate_at function for the sake of readability, interfacing with databases, etc.

Below are some similar questions asked on stackoverflow which do not address the question I posed here:

adding multiple columns in a dplyr mutate call

dplyr::mutate to add multiple values

Use of column inside sum() function using dplyr's mutate() function

Answer 1

This was answered by @eipi10 in @eipi10's comment on the question, but I'm writing it here for posterity.

The solution here is to use:

df %>%
   mutate_at(.vars = vars(y, z),
             .funs = list(~ ifelse(x, ., NA)))

You can also use the new across() function with mutate() , like so:

df %>%
   mutate(across(c(y, z), ~ ifelse(x, ., NA)))

The use of the formula operator (as in ~ ifelse(...) ) here indicates that ifelse(x, ., NA) is an anonymous function that is being defined within the call to mutate_at() .

This works similarly to defining the function outside of the call to mutate_at() , like so:

temp_fn <- function(input) ifelse(test = df[["x"]],
                                  yes = input,
                                  no = NA)

df %>%
   mutate_at(.vars = vars(y, z),
             .funs = temp_fn)

Note on syntax changes in dplyr: Prior to dplyr version 0.8.0, you would simply write .funs = funs(ifelse(x, . , NA)) , but the funs() function is being deprecated and will soon be removed from dplyr.

Answer 2

To supplement the previous response, if you wanted mutate_at() to add new variables (instead of replacing), with names such as z_1 and y_1 as in the original question, you just need to:

• dplyr >=1 with across() : add .names="{.col}_1" , or alternatively use list('1'=~ifelse(x, ., NA) (back ticks!)
• dplyr [0.8, 1[ : use list('1'=~ifelse(x, ., NA)
• dplyr <0.8 : use funs('1'=ifelse(x, ., NA)

library(tidyverse)

df <- data.frame(
  x = c(TRUE, TRUE, FALSE),
  y = c("Hello", "Hola", "Ciao"),
  z = c("World", "ao", "HaOlam")
)

## Version >=1
df %>%
  mutate(across(c(y, z), 
                list(~ifelse(x, ., NA)),
                .names="{.col}_1"))
#>       x     y      z   y_1   z_1
#> 1  TRUE Hello  World Hello World
#> 2  TRUE  Hola     ao  Hola    ao
#> 3 FALSE  Ciao HaOlam  <NA>  <NA>


## 0.8 - <1
df %>%
  mutate_at(.vars = vars(y, z),
            .funs = list(`1`=~ifelse(x, ., NA)))
#>       x     y      z   y_1   z_1
#> 1  TRUE Hello  World Hello World
#> 2  TRUE  Hola     ao  Hola    ao
#> 3 FALSE  Ciao HaOlam  <NA>  <NA>

## Before 0.8
df %>%
  mutate_at(.vars = vars(y, z),
            .funs = funs(`1`=ifelse(x, ., NA)))
#> Warning: `funs()` is deprecated as of dplyr 0.8.0.
#> Please use a list of either functions or lambdas: 
#> 
#>   # Simple named list: 
#>   list(mean = mean, median = median)
#> 
#>   # Auto named with `tibble::lst()`: 
#>   tibble::lst(mean, median)
#> 
#>   # Using lambdas
#>   list(~ mean(., trim = .2), ~ median(., na.rm = TRUE))
#> This warning is displayed once every 8 hours.
#> Call `lifecycle::last_warnings()` to see where this warning was generated.
#>       x     y      z   y_1   z_1
#> 1  TRUE Hello  World Hello World
#> 2  TRUE  Hola     ao  Hola    ao
#> 3 FALSE  Ciao HaOlam  <NA>  <NA>

^{Created on 2020-10-03 by the reprex package (v0.3.0)}

For more details and tricks, see: Create new variables with mutate_at while keeping the original ones