Select column with the longest string in R

user_13647125 0
Viewed 842

I have this type of dataset

DF    
     V1     V2     V3
1.   A      AAA    B
2.   BBB    B      CC
3.   C      BB     CCC

And I would like to select the longest string from DF and put it into the new column WINNER like this:

DF    
     V1     V2     V3    WINNER
1.   A      AAA    B     AAA
2.   BBB    B      CC    BBB
3.   C      BB     CCC   CCC

I have tried

mutate( WINNER = select(which.max (c(nchar(V1), nchar(V2), nchar(V3)))

but it works only for numeric values. I would prefer dplyr solution.

dplyr r select comparison string
6 Answers
df$winner <- apply(df,1, function(x) x[which.max(nchar(x))])

df
    V1  V2  V3 winner
1.   A AAA   B    AAA
2. BBB   B  CC    BBB
3.   C  BB CCC    CCC
user_104424230

In case of of ties winner will be based on first apperance:

df$WINNER <- apply(df, 1, function(row) row[which.max(nchar(row))])
user_121484020

You can use c_across(). What you put in there will control which columns are selected.

library(dplyr)

df %>% 
  rowwise() %>% 
  mutate(WINNER = c_across(starts_with("V"))[which.max(nchar(c_across(starts_with("V"))))])

It can be a bit more compact if you want all columns.

df %>% 
  rowwise() %>% 
  mutate(WINNER = c_across()[which.max(nchar(c_across()))])
user_109174790 
df$winner <- 
  Reduce(function(x, y) ifelse(nchar(y) > nchar(x), y, x), df) 

df
#     V1  V2  V3 winner
# 1:   A AAA   B    AAA
# 2: BBB   B  CC    BBB
# 3:   C  BB CCC    CCC
user_67627550

One dplyr option could be.

df %>%
 rowwise() %>%
 mutate(WINNER = get(paste0("V", which.max(nchar(c_across(V1:V3))))))

  V1    V2    V3    WINNER
  <chr> <chr> <chr> <chr> 
1 A     AAA   B     AAA   
2 BBB   B     CC    BBB   
3 C     BB    CCC   CCC
user_59645570

It's a shame that max/min in R doesn't have the key argument like in python, but one can quickly cook up something similar. I would suggest something like this:

library(tidyverse)


df <- read_table(
  "
  V1     V2     V3
  A      AAA    B
  BBB    B      CC
  C      BB     CCC
  "
)

max_key <- function(vars, fn) {
  vars[which.max(fn(vars))]
}

df %>% 
  rowwise() %>% 
  mutate(
    winner = max_key(c_across(V1:V3), str_length)
  )
#> # A tibble: 3 x 4
#> # Rowwise: 
#>   V1    V2    V3    winner
#>   <chr> <chr> <chr> <chr> 
#> 1 A     AAA   B     AAA   
#> 2 BBB   B     CC    BBB   
#> 3 C     BB    CCC   CCC

Created on 2020-06-26 by the reprex package (v0.3.0)

user_77787870
Related