Converting Column Names into Values in New Columns with Conditions| R-CodePudding

This data set has the name of the questions as columns and scores as the values. But candidates will only answer (2) questions so there are alot of NULL values.

I am trying to make (2) new columns [Question#1] & [Question #2] to show which question the candidate took when there is a score present in the column.

I haven't been successful in trying to make code to run this smoothly. It would be filtering out "N/A" and matching with a number to call back the column name.

Example Output:

DPUT:

    structure(list(candidate_id = c(1, 2, 3, 4, 5, 6), `Office Design` = c("N/A", 
    "N/A", "50", "N/A", "N/A", "N/A"), `Better Compression` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Simple Customer Support` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Break a Palindrome` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Delete Nnodes Greater Than X` = c("N/A", 
    "0", "N/A", "N/A", "50", "N/A"), `Double Size` = c("N/A", "N/A", 
    "N/A", "N/A", "N/A", "N/A"), `Game Winner` = c("N/A", "N/A", 
    "N/A", "N/A", "N/A", "50"), `Hashed Ports` = c("N/A", "N/A", 
    "N/A", "N/A", "N/A", "N/A"), `Maximum Substring` = c("N/A", "N/A", 
    "N/A", "50", "N/A", "N/A"), `Packet Stream Player` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Separate the Files` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Subarray Sum` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Tag Identification Number` = c("0", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Threshold Alerts` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Array Journey` = c("N/A", 
    "N/A", "N/A", "N/A", "75", "N/A"), `Barter Market` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), Encircular = c("N/A", "N/A", 
    "N/A", "N/A", "N/A", "N/A"), `Find the Substring` = c("N/A", 
    "N/A", "75", "N/A", "N/A", "N/A"), `Perfect Substring` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Reductor Array` = c("N/A", 
    "0", "N/A", "9", "N/A", "N/A"), `Shortest Palindrome` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Similar Numbers` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Subarray Products` = c("0", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Minimum Start Value` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "N/A"), `Website Pagination` = c("N/A", 
    "N/A", "N/A", "N/A", "N/A", "48")), class = c("tbl_df", "tbl", 
    "data.frame"), row.names = c(NA, -6L))

CodePudding user response：

You can pivot this to long format, change the values to numeric, and filter those that are not NA, make a row id using row_number(), and then pivot back to wide format.

qs = pivot_longer(df, -1) %>% 
  mutate(value=as.numeric(value)) %>%
  filter(!is.na(value)) %>% 
  group_by(candidate_id) %>% 
  mutate(row=row_number()) %>% 
  ungroup() %>% 
  pivot_wider(id_cols = candidate_id, names_from=row, values_from = name,names_prefix = "Question_")

qs looks like this:

  candidate_id Question_1                   Question_2        
         <dbl> <chr>                        <chr>             
1            1 Tag Identification Number    Subarray Products 
2            2 Delete Nnodes Greater Than X Reductor Array    
3            3 Office Design                Find the Substring
4            4 Maximum Substring            Reductor Array    
5            5 Delete Nnodes Greater Than X Array Journey     
6            6 Game Winner                  Website Pagination

You can keep it like that, or if you want, join it back to the original, like this:

inner_join(df, qs)

A more concise and a faster approach uses data.table:

library(data.table)
qs = dcast(
  melt(setDT(df),id="candidate_id")[value!="N/A"][, id:=paste0("Question_",1:.N), candidate_id],
  candidate_id~id, value.var="variable"
)

df[qs, on =.(candidate_id)]