Convert various dummy/logical variables into a single categorical variable/factor from their name in R

Question

My question has strong similarities with this one and this other one, but my dataset is a little bit different and I can't seem to make those solutions work. Please excuse me if I misunderstood something and this question is redundant.

I have a dataset such as this one:

df <- data.frame(
  id = c(1:5),
  conditionA = c(1, NA, NA, NA, 1),
  conditionB = c(NA, 1, NA, NA, NA),
  conditionC = c(NA, NA, 1, NA, NA),
  conditionD = c(NA, NA, NA, 1, NA)
  )
# id conditionA conditionB conditionC conditionD
# 1  1          1         NA         NA         NA
# 2  2         NA          1         NA         NA
# 3  3         NA         NA          1         NA
# 4  4         NA         NA         NA          1
# 5  5          1         NA         NA         NA

(Note that apart from these columns, I have a lot of other columns that shouldn't be affected by the current manipulation.)

So, I observe that conditionA, conditionB, conditionC and conditionD are mutually exclusives and should be better presented as a single categorical variable, i.e. factor, that should look like this :

#   id       type
# 1  1 conditionA
# 2  2 conditionB
# 3  3 conditionC
# 4  4 conditionD
# 5  5 conditionA

---

I have investigated using gather or unite from tidyr, but it doesn't correspond to this case (with unite, we lose the information from the variable name).

I tried using kimisc::coalescence.na, as suggested in the first referred answer, but 1. I need first to set a factor value based on the name for each column, 2. it doesn't work as expected, only including the first column :

library(kimisc)
# first, factor each condition with a specific label
df$conditionA <- df$conditionA %>%
  factor(levels = 1, labels = "conditionA")
df$conditionB <- df$conditionB %>%
  factor(levels = 1, labels = "conditionB")
df$conditionC <- df$conditionC %>%
  factor(levels = 1, labels = "conditionC")
df$conditionD <- df$conditionD %>%
  factor(levels = 1, labels = "conditionD")

# now coalesce.na to merge into a single variable
df$type <- coalesce.na(df$conditionA, df$conditionB, df$conditionC, df$conditionD)

df
#   id conditionA conditionB conditionC conditionD       type
# 1  1 conditionA       <NA>       <NA>       <NA> conditionA 
# 2  2       <NA> conditionB       <NA>       <NA>       <NA> 
# 3  3       <NA>       <NA> conditionC       <NA>       <NA> 
# 4  4       <NA>       <NA>       <NA> conditionD       <NA> 
# 5  5 conditionA       <NA>       <NA>       <NA> conditionA

I tried the other suggestions from the second question, but haven't found one that would bring me the expected result...

Answer 1

Try:

library(dplyr)
library(tidyr)

df %>% gather(type, value, -id) %>% na.omit() %>% select(-value) %>% arrange(id)

Which gives:

#  id       type
#1  1 conditionA
#2  2 conditionB
#3  3 conditionC
#4  4 conditionD
#5  5 conditionA

---

Update

To handle the case you detailed in the comments, you could do the operation on the desired portion of the data frame and then left_join() the other columns:

df %>% 
  select(starts_with("condition"), id) %>% 
  gather(type, value, -id) %>% 
  na.omit() %>% 
  select(-value) %>% 
  left_join(., df %>% select(-starts_with("condition"))) %>%
  arrange(id)