Put column name as a value across several columns, under condition - r

Question

I have a dataset looking like that:
set.seed(123)
test_data <- data.frame(
id = c("a", "b", "c", "d", "e"),
x = sample(c(0,1), 5, replace = T),
y = sample(c(0,1), 5, replace = T)
)
> test_data
id x y
1 a 0 1
2 b 0 1
3 c 0 1
4 d 1 0
5 e 0 0
For the columns x and y, if the value is equal to 1, the value is replaced by the name of the column. In my example, I would like to have:
id x y
1 a <NA> y
2 b <NA> y
3 c <NA> y
4 d x <NA>
5 e <NA> <NA>
The thing that I don't know how many columns should be treated this way. Basically, I know that the first column ("id") is not concerned, but after this column, I could have any number of columns (even 0) that need to be treated this way.
I tried something like that but it doesn't work:
library(dplyr)
test_data %>%
mutate(
across(
.cols = 1:last_col(),
.funs = function(x) {
ifelse(x == 1, as.character(x), NA)
}
)
)
How can I do that? A dplyr answer is preferred.

Answer 1

You can do this way also:
library(tidyverse)
## define a function for your job
fn <- function(x, name){
return(ifelse(x ==1, name, NA))
}
test_data %>%
select(-id) %>%
map2_dfr(., names(.), ~fn(.x, .y)) %>%
bind_cols('id'= test_data$id, .)
Another version could be:
fn <- function(x, name){
if(name != 'id'){
return(ifelse(x ==1, name, NA))
} else {
return(x)
}
}
test_data %>%
map2_dfr(., names(.), ~fn(.x, .y))

Answer 2

Here is a try using purrr
library(dplyr)
library(purrr)
col_to_fix <- names(test_data)[2:length(test_data)]
walk(.x = col_to_fix, .f = function(x) {
# Note that I used <<- assigment here to change the test_data in global
test_data[[x]] <<- case_when(
test_data[[x]] == 1 ~ x,
TRUE ~ NA_character_
)
})
Output
> test_data
id x y
1 a <NA> y
2 b <NA> y
3 c <NA> y
4 d x <NA>
5 e <NA> <NA>