Can we avoid repeating the column name each time when using it in mutates

Question

I use several mutate on the same column. How can we only use mutate once and without repeating the column name?

df <- data.frame(
  c1 = c("Élève", "Café", "Château", "Noël", "Crème")
)

df2 <- df %>% 
  mutate(c1 = trimws(c1)) %>%
  mutate(c1 = gsub("\\s+", " ", c1)) %>%
  mutate(c1 = gsub("\"", "", c1)) %>%
  mutate(c1 = iconv(toupper(c1), to = "ASCII//TRANSLIT"))

score 7 · Accepted Answer · answered Aug 21 '23 at 13:22

7

Place the pipeline within the mutate like this:

df3 <- df %>%
  mutate(c1 = c1 %>%
    trimws %>%
    gsub("\\s+", " ", .) %>%
    gsub("\"", "", .) %>%
    toupper %>%
    iconv(to = "ASCII//TRANSLIT"))

identical(df2, df3)
## [1] TRUE

answered Aug 21 '23 at 13:22

G. Grothendieck

254,981
17
203
341

1

https://style.tidyverse.org/pipes.html#no-arguments style guide hint ;) – Mark Aug 21 '23 at 13:24
1

The style guide is just someone's opinion. If you like it go with it but if not just ignore it. In particular I would not use `trimws()` when the shorter `trimws` is supported. – G. Grothendieck Aug 21 '23 at 13:25
I mean, so is spelling, but point tayken – Mark Aug 21 '23 at 13:26

Mark · Answer 2 · 2023-08-21T13:45:41.970

You can use pipes within mutate calls! Also, even if that weren't the case, columns you create in a mutate function call can be used later within the same function call. So you could keep on redefining c1 within one mutate call.

But anyway, this is how I would do it (using almost all stringr functions):

library(stringr)

df2 <- df |>
  mutate(c1 = str_squish(c1) |>
              str_remove_all("\"") |>
              str_to_upper() |>
              iconv(to = "ASCII//TRANSLIT"))

score 2 · Answer 3 · answered Aug 21 '23 at 14:42

Not that you need another solution, but it could be handy to combine all your steps into a single function to tidy up your mutate call. You can combine a string of functions easily with purrr::compose to run them in the given order each time you need them.

Using G. Grothendieck's excellent code split into anonymous functions:

library(tidyverse)

df <- data.frame(
  c1 = c("Élève", "Café", "Château", "Noël", "Crème")
)

tidy_text <- compose(
  \(t) gsub("\\s+", " ", t),
  \(t) gsub("\"", "", t),
  toupper,
  \(t) iconv(t, to = "ASCII//TRANSLIT")
)

df %>% 
  mutate(c1 = tidy_text(c1))
#>        c1
#> 1   ELEVE
#> 2    CAFE
#> 3 CHATEAU
#> 4    NOEL
#> 5   CREME

Or using Mark's tidyverse code and purrr formula/function syntax:

tidy_text2 <- compose(
  str_squish,
  ~ str_remove_all(.x, "\""),
  str_to_upper,
  ~ iconv(.x, to = "ASCII//TRANSLIT")
)

df %>%
  mutate(c1 = tidy_text2(c1))
#>        c1
#> 1   ELEVE
#> 2    CAFE
#> 3 CHATEAU
#> 4    NOEL
#> 5   CREME

May not be necessary if you're only using it once of course! Just one way of having some bits tidier than others!

Can we avoid repeating the column name each time when using it in mutates

3 Answers3