R: Get column names where value is not null - r

Question

I have a table of 7 columns, the first column is id, then 3 columns of vegetable types and the last 3 columns are fruit types. The values indicate whether a person has this vegetable/ fruit. Is there a way to group the vegetables and the fruits, and output the column names if the person has that vegetable/ fruit?
Input data frame:
id1 <- c("id_1", 1, NA, NA, NA, 1, NA)
id2 <- c("id_2", NA, 1, 1, NA, NA, NA)
input <- data.frame(rbind(id1, id2))
colnames(input) = c("id", "lettuce", "tomato", "bellpeper", "pineapple", "apple", "banana")
Expected output data frame:
output_id1 <- c("id_1", "lettuce", "apple")
output_id2 <- c("id_2", "tomato, bellpeper", NA)
output <- data.frame(rbind(output_id1, output_id2))
colnames(output) <- c("id", "veg", "fruit")

Answer 1

Using the original input data you posted (also shown below in Data) you could do this with the tidyr package:
library(tidyr)
input %>%
tidyr::pivot_longer(cols = matches("^veg|^fruit"),
names_sep = "_",
names_to = c("type", "val"),
values_drop_na = T) %>%
tidyr::pivot_wider(id_cols = id,
names_from = type,
values_from = val,
values_fn = function(x) paste0(x, collapse = ","))
Output
id veg fruit
<chr> <chr> <chr>
1 id_1 lettuce apple
2 id_2 tomato,bellpeper NA
Data
input <- structure(list(id = c("id_1", "id_2"), veg_lettuce = c("1", NA
), veg_tomato = c(NA, "1"), veg_bellpeper = c(NA, "1"), fruit_pineapple = c(NA_character_,
NA_character_), fruit_apple = c("1", NA), fruit_banana = c(NA_character_,
NA_character_)), class = "data.frame", row.names = c("id1", "id2"
))

Answer 2

This should do the trick!
id1 <- c("id_1", 1, NA, NA, NA, 1, NA)
id2 <- c("id_2", NA, 1, 1, 1, NA, NA)
input <- data.frame(rbind(id1, id2))
colnames(input) = c("id", "lettuce", "tomato", "bellpeper", "pineapple", "apple", "banana")
# Remove the id column, it's not necessary
input_without_id <- dplyr::select(input, -c("id"))
# For each row (margin = 1) of the input, return the names vector (names(input))
# but only in the positions the where the row (x!) is not NA
result <- apply(input_without_id, MARGIN = 1, function(x) {
return(names(input_without_id)[which(!is.na(x))])
})
# Rename the result with the corresponding ids originally found in input.
names(result) <- input$id

Answer 3

Here is a tidyverse solution:
library(tidyverse)
input %>%
pivot_longer(-id) %>%
group_by(id) %>%
separate(name, into = c('type', 'class'), sep = "_") %>%
na.omit() %>%
select(-value) %>%
group_by(id, type) %>%
summarise(class = toString(class)) %>%
ungroup() %>%
pivot_wider(names_from = type, values_from = class) %>%
unnest() %>%
select(id, veg, fruit)
This gives us:
# A tibble: 2 x 3
id veg fruit
<chr> <chr> <chr>
1 id_1 lettuce apple
2 id_2 tomato, bellpeper NA