Using and pasting current column name into mutate case_when

Question

My data looks like this:

dt <- structure(list(var1_dummy = c(0, 1, 0, 0, 0, 1, 0, 0, 1, 1), 
    var2_dummy = c(1, 0, 0, 0, 0, 1, 0, 0, 1, 1), var1_scale = c(NA, 
    3, NA, NA, NA, 3, NA, NA, 4, 4), var2_scale = c(3, NA, NA, 
    NA, NA, 2, NA, NA, 3, 5)), class = "data.frame", row.names = c(NA, 
-10L))

   var1_dummy var2_dummy var1_scale var2_scale
1           0          1         NA          3
2           1          0          3         NA
3           0          0         NA         NA
4           0          0         NA         NA
5           0          0         NA         NA
6           1          1          3          2
7           0          0         NA         NA
8           0          0         NA         NA
9           1          1          4          3
10          1          1          4          5

I now want to mutate the variables with suffix "scale" in a case_when that evaluates the corresponding variable with suffix "dummy". (So manipulation of var1_scale should depend on var1_dummy etc.) The new version of var1-scale should be 0 if var1_dummy is 0, and should augment by 1 if var1_dummy is 1.

Note that I have many of such columns, so mutating every column individually should be avoided.

The variables to be mutated are in the following vector:

vars <- v(var1_scale, var2_scale)

Now I manage to do what I want with the good ol' loop:

for (var in vars) {
  dummy <- gsub("scale", "dummy", var)
  dt[, outlet] <- case_when(
    dt[[outlet_expo]] == 0 ~ 0,
    dt[[outlet_expo]] == 1 ~ dt[[outlet]] + 1)
}

However, I'd prefer a vectorised solution. Here's what I tried:

dt %>% 
  mutate(across(all_of(vars),
                ~ case_when(
                  !!as.symbol(gsub("scale", "dummy", as.name(cur_column()))) == 0 ~ 0,
                  !!as.symbol(gsub("scale", "dummy", as.name(cur_column()))) == 1 ~ . + 1))

... the idea being that I take the name of the current column, change it with gsub and then evaluate it as a column again. But cur_column does not seem to work inside a case-when.

Answer 1

with data.table you can do something like:

dummy_vars <- names(dt)[grep("dummy", names(dt))] 
scale_vars <- names(dt)[grep("scale", names(dt))] 
setDT(dt)[, (scale_vars) := map2(mget(dummy_vars), mget(scale_vars), ~ifelse(.x == 0, .x, .y + 1))]

or easier using the fact that when dummy col is 0 then scale one is NA:

scale_vars <- names(dt)[grep("scale", names(dt))] 
setDT(dt)[, (scale_vars) := map(.SD, ~ifelse(is.na(.x), 0, .x + 1)), .SDcols = scale_vars]

Answer 2

a bit late, but this might help:

library(tidyverse)
dt %>% 
  mutate(across(ends_with("_scale"),
                ~case_when(
                  cur_data() %>% select(all_of(gsub("scale","dummy",cur_column()))) == 0 ~ 0,
                  cur_data() %>% select(all_of(gsub("scale","dummy",cur_column()))) == 1 ~ . + 1)
  ))

Rather than using as.symbol, I'd simply use cur_data() and then select the dummy column from it again using a gsub() within the select() function.

Also remember that cur_column() already gives you a character vector of the name of the column.

You could consider adding a pull() to the end to really make sure you have a vector, but I think it works without it as well.

Hope this helps!

Answer 3

You can add two similar sized dataframes directly, just change the NA values to 0 in scale columns.

dummy_cols <- grep('dummy', names(dt))
scale_cols <- grep('scale', names(dt))

dt[scale_cols] <- dt[dummy_cols] + replace(dt[scale_cols], is.na(dt[scale_cols]), 0)
dt

#   var1_dummy var2_dummy var1_scale var2_scale
#1           0          1          0          4
#2           1          0          4          0
#3           0          0          0          0
#4           0          0          0          0
#5           0          0          0          0
#6           1          1          4          3
#7           0          0          0          0
#8           0          0          0          0
#9           1          1          5          4
#10          1          1          5          6