How do I select a respondent's value if conditions are met by row?

Question

I have a dataset that looks like this. For respondents who only picked one favorite fruit, I need to have variable returned that is a factor which is that respondent's favorite fruit. If a respondent picked more than one favorite fruit, their value should be NA.

fav_apple<-sample(c(NA,1), size=100, replace=T, prob=c(0.9,0.1))
fav_orange<-sample(c(NA,1), size=100, replace=T, prob=c(0.9,0.1))
fav_banana<-sample(c(NA,1), size=100, replace=T, prob=c(0.9,0.1))
df<-data.frame(fav_apple, fav_orange, fav_banana)
df$id<-seq(1,nrow(df))

I tried tackling this with pivot_longer and then repivotting, but I end up with duplicate rows.

Answer 1

Using ifelse on the rowSums. Since favorites is a numeric matrix apply can be used efficiently on as.matrix(df[,1:3])

colnms <- colnames(df[,1:3])

cbind(df, 
  favorite = as.factor(unlist(
    ifelse(rowSums(df[,1:3], na.rm=T) == 1, 
           apply(as.matrix(df[,1:3]), 1, \(x) colnms[which(!is.na(x))]), 
           NA))))
    fav_apple fav_orange fav_banana  id   favorite
1          NA         NA         NA   1       <NA>
2          NA         NA         NA   2       <NA>
3          NA          1         NA   3 fav_orange
4          NA         NA         NA   4       <NA>
5           1          1          1   5       <NA>
6          NA         NA         NA   6       <NA>
7          NA         NA         NA   7       <NA>
8          NA         NA         NA   8       <NA>
9          NA         NA         NA   9       <NA>
10         NA         NA         NA  10       <NA>
11         NA         NA         NA  11       <NA>
12         NA         NA         NA  12       <NA>
13         NA         NA         NA  13       <NA>
14         NA         NA         NA  14       <NA>
15         NA         NA          1  15 fav_banana
...