I have a dataframe like:

I would like to replace all non-NA cells (i.e. that have value: a, b, c, d, x, y, z) with the header names:

Thanks!

CodePudding user response：

Using cur_column() inside ifelse and across:

library(tidyverse)

df = tibble(
  header1 = sample(c(NA, "a", "b", "c"), 4),
  header2 = sample(c(NA, "a", "b", "c"), 4),
  header3 = sample(c(NA, "a", "b", "c"), 4),
  header4 = sample(c(NA, "a", "b", "c"), 4))

> df
# A tibble: 4 × 4
  header1 header2 header3 header4
  <chr>   <chr>   <chr>   <chr>  
1 b       a       NA      b      
2 NA      NA      c       a      
3 a       c       a       c      
4 c       b       b       NA   

> df %>% mutate(across(.fns = ~ ifelse(!is.na(.), cur_column(), NA)))
# A tibble: 4 × 4
  header1 header2 header3 header4
  <chr>   <chr>   <chr>   <chr>  
1 header1 header2 NA      header4
2 NA      NA      header3 header4
3 header1 header2 header3 header4
4 header1 header2 header3 NA  

CodePudding user response：

Ricardo Semião e Castro's answer works perfectly well using tidyverse package and syntax, but you can use base R as well by for-looping over the columns:

df = data.frame(
  header1 = sample(c(NA, "a", "b", "c"), 4),
  header2 = sample(c(NA, "a", "b", "c"), 4),
  header3 = sample(c(NA, "a", "b", "c"), 4),
  header4 = sample(c(NA, "a", "b", "c"), 4))

colnames <- colnames(df)
for(col in colnames){
  df[[col]][!is.na(df[[col]])] <- col
}

CodePudding user response：

You can also do it in one assignment in base R without a loop:

df[!is.na(df)] <- names(df)[col(df)[!is.na(df)]]