How to map custom function across multiple variables-CodePudding

I wrote a function to determine whether for a given variable, the values in a group are not all the same and create a new variable that provides the name of the variable if there is a difference or NA if there is none.

x <- c(2,4,5,5,6,2,3)
y <- c(5,5,2,3,6,1,8)
z <- c(5,2,4,1,3,5,1)
xy <- tibble(x, y, z)

diff_var <- function(a, b) {
 a %>% 
    transmute("{{c}}" := n_distinct({{c}})) %>% 
    ungroup() %>% 
    select({{c}}) %>% 
    imap_dfc(~ if_else(.x > 1, .y, NA_character_))
}

xy %>% 
  group_by(x) %>% 
  diff_var(., y)

# A tibble: 7 × 1
  y    
  <chr>
1 y    
2 NA   
3 y    
4 y    
5 NA   
6 y    
7 NA

I’m now trying to figure out how to do this across multiple variables (ideally excluding the grouping variable). For the sample data here, y and z. The variable to group by will be the same for all variables. My various attempt at using different forms of map have failed; I'm struggling to get the arguments into my custom function accurately. Eventually I'll want to have the grouping variable included but I can easily add that back later.

Desired output:

# A tibble: 7 × 2
  y     z    
  <chr> <chr>
1 y     NA   
2 NA    NA   
3 y     z    
4 y     z    
5 NA    NA   
6 y     NA   
7 NA    NA

Though there are similarly titled questions on this site, I haven't been able to adapt them to my current case.

CodePudding user response：

It may be easier to use across and pass a vector of column names in c as quoted. In addition, the last step can be done within dplyr itself i.e. using across

diff_var <- function(a, c) {
 a %>% 
    transmute(across(all_of(c), n_distinct)) %>% 
    ungroup() %>% 
    select(all_of(c)) %>% 
    mutate(across(everything(), ~ case_when(.x > 1~ cur_column())))
}

-testing

xy %>% 
   group_by(x) %>% 
   diff_var(., c("y", "z"))
# A tibble: 7 × 2
  y     z    
  <chr> <chr>
1 y     <NA> 
2 <NA>  <NA> 
3 y     z    
4 y     z    
5 <NA>  <NA> 
6 y     <NA> 
7 <NA>  <NA>