I have the following dataframe.
library(dplyr)
data_test = data.frame(va_te=c("yes", "", "no", "yes"),
va_ti=c("no", "", "yes", "no"),
va_ze=c("", "no", "yes", "no"),
jk_te=c(NA, 545, 876, 987),
jk_ti=c(876, 567, 908, 432),
jk_ze=c(987, 988, NA, 234),
loc=c(345, 898, 444, 321))
> data_test
va_te va_ti va_ze jk_te jk_ti jk_ze loc
1 yes no NA 876 987 345
2 no 545 567 988 898
3 no yes yes 876 908 NA 444
4 yes no no 987 432 234 321
I would like to mutate the first 6 variables using two patterns (one for each group of 3 similar variables: va_ and jk_).
For the first group "va_" I used across + starts_with + ifelse.
For the second group "jk_" I would like to use the same method but as my ifelse uses both jk and va in the test, I don't know how to write this part.
data_test_2 = data_test %>%
mutate(across(starts_with("va_"),
~ ifelse(.x == "", NA, .x), .names = "{col}")) %>%
mutate(jk_te = ifelse(va_te != "yes", NA,
ifelse(va_te == "yes" & is.na(jk_te), loc, jk_te))) %>%
mutate(jk_ti = ifelse(va_ti != "yes", NA,
ifelse(va_ti == "yes" & is.na(jk_ti), loc, jk_ti))) %>%
mutate(jk_ze = ifelse(va_ze != "yes", NA,
ifelse(va_ze == "yes" & is.na(jk_ze), loc, jk_ze)))
> data_test_2
va_te va_ti va_ze jk_te jk_ti jk_ze loc
1 yes no <NA> 345 NA NA 345
2 <NA> <NA> no NA NA NA 898
3 no yes yes NA 908 444 444
4 yes no no 987 NA NA 321
You can use sub "jk" with "va" on the name of the current column (cur_column()), and get the columns for the actions.
library(dplyr)
data_test %>%
mutate(across(starts_with("va_"), ~ ifelse(.x == "", NA, .x)),
across(starts_with("jk_"), ~ ifelse(get(sub("jk", "va", cur_column())) != "yes", NA,
ifelse(get(sub("jk", "va", cur_column())) == "yes" & is.na(.x), loc, .x))))
va_te va_ti va_ze jk_te jk_ti jk_ze loc
1 yes no <NA> 345 NA NA 345
2 <NA> <NA> no NA NA NA 898
3 no yes yes NA 908 444 444
4 yes no no 987 NA NA 321