Applying dplyr's rename to all columns while using pipe operator

Question

I'm working with an imported data set that corresponds to the extract below:

set.seed(1)
dta <- data.frame("This is Column One" = runif(n = 10),
                     "Another amazing Column name" = runif(n = 10),
                     "!## This Columns is so special€€€" = runif(n = 10),
                    check.names = FALSE)

I'm doing some cleaning on this data using dplyr and I would like to change column names to syntatically correct ones and remove the punctuation as a second step. What I tried so far:

dta_cln <- dta %>% 
    rename(make.names(names(dta)))

generates an error:

> dta_clean <- dta %>% 
+     rename(make.names(names(dta)))
Error: All arguments to rename must be named.

Desired result

What I wan to achieve can be done in base:

names(dta) <- gsub("[[:punct:]]","",make.names(names(dta)))

which would return:

> names(dta)
[1] "ThisisColumnOne"          "AnotheramazingColumnname" "XThisColumnsissospecial"

I want to achieve the same effect but using dyplr and %>%.

Looks like some tweaking of [this](http://stackoverflow.com/questions/30382908/r-dplyr-rename-variables-using-string-functions) — akrun, Dec 04 '15 at 16:07
@akrun Thanks very much, I will try to do something with `setNames(tolower(gsub("\\.","_",names(.))))` as suggested in the linked answer. — Konrad, Dec 04 '15 at 16:09
Only problem is that some characters are not parsing well within the `rename`. — akrun, Dec 04 '15 at 16:10
Yup: `Error in parse(text = x) : :1:9: unexpected symbol 1: Service Condiitions` — Konrad, Dec 04 '15 at 16:11
After tweaking, [this](http://stackoverflow.com/a/30383251/1655567) will work. — Konrad, Dec 04 '15 at 16:14
Yes, the standalone should work outside the `rename`. But, I understand the reason for making it right with the `dplyr` functions itself. — akrun, Dec 04 '15 at 16:15
For some reason I've taken to doing this if I know I'm going to rename everything: `iris %T>% { colnames(.) <- paste0("iris_",names(.)) }`. Somewhat unclear to me why I prefer it to `rename`. — Akhil Nair, Jul 13 '18 at 09:23

score 35 · Accepted Answer · answered Aug 10 '17 at 19:05

I know this is an old question, and I'm sure you found the solution by now, but I stumbled here searching for the same question, and ultimately found a few new ways to do this.

Dplyr

Using dplyr 0.6.0 and above, there is now a rename_all function:

  dta %>% 
    rename_all(funs(gsub("[[:punct:]]", "", make.names(names(dta)))))

Which works, but it's a little messy to me. If you want more flexibility with dplyr, you can also call on:

rename_at
rename_if

Janitor

This is a pretty nice package (with plenty of additional utility) that can easily clean up column names:

library(janitor)

dta %>% 
  clean_names()

Which will rename and clean all column names to the following:

[1] "this_is_column_one"  "another_amazing_column_name"  "x_this_columns_is_so_special"

Everything becomes snake_case rather than CamelCase, but overall clean_names is very flexible in the column names it handles. If that IS a deal breaker, you can use yet another package snakecase for its function to_big_camel_case() within the rename_all function...although that is starting to get a little too esoteric

`funs()` is deprecated as of dplyr 0.8.0. Looks like you now want: `dta %>% rename_all(list(~ gsub("[[:punct:]]", "", .)))` or (since `rename_all()` has been superceded by `rename_with()` ... `dta %>% rename_with(~ gsub("[[:punct:]]", "", .x))` — Brian D, Feb 17 '21 at 19:16

score 34 · Answer 2 · answered May 25 '19 at 10:44

Set column names with the pipe like so:

iris %>% `colnames<-`(c("newcol1", "newcol2", "newcol3", "newcol4", "newcol5"))

Which returns

    newcol1 newcol2 newcol3 newcol4    newcol5
1       5.1     3.5     1.4     0.2     setosa
2       4.9     3.0     1.4     0.2     setosa
3       4.7     3.2     1.3     0.2     setosa

Jiaxiang · Answer 3 · 2018-05-17T03:25:23.977

6

mtcars %>% 
  data.table::setnames(
    old = mtcars %>% names(),
    new = mtcars %>% names() %>% paste0("_new_name")
  )

The function setnames in data.table package is to rename the column names in data frame. old and new are two arguments in this function we need.

mtcars %>% names() outputs the column names of data frame mtcars in pipeline %>% way, so you can also use names(mtcars). They are same thing.

In this minimal example, I rename the column names in pipeline %>% and add all old column names with a postfix using paste0 function. You can add prefix, postfix or other rules.

edited May 17 '18 at 03:25

answered Apr 30 '18 at 13:39

Jiaxiang

865
12
23

Please add some explanation to your answer. Why is your answer better than the accepted answer for example? – Jesse Apr 30 '18 at 14:05

score 3 · Answer 4 · answered Oct 13 '17 at 19:16

You can also try this

set.seed(1)
dta <- data.frame("This is Column One" = runif(n = 10),
                 "Another amazing Column name" = runif(n = 10),
                 "!## This Columns is so special€€€" = runif(n = 10),
                check.names = FALSE)

dta <- dta  %>% 
  setNames(gsub("[^[:alnum:] ]", perl = TRUE,
            "",
            names(.))) %>% 
  setNames(gsub("(\\w)(\\w*)",
            "\\U\\1\\L\\2",
            perl = TRUE,
            names(.)))

names(dta)
[1] "This Is Column One"          "Another Amazing Column Name" " This Columns Is So Special"

This should be the accepted answer. The others depend on the dataframe being assigned in the first place in order to then alter colnames. Thanks for this! — Anurag N. Sharma, Jun 29 '20 at 08:57

score 1 · Answer 5 · answered Nov 29 '19 at 09:26

1

Using Stringr and Dplyr, and the dot operator:

dta %>%
   dplyr::rename_all(funs(
                     stringr::str_replace_all( ., "[[:punct:]]", "_" )
   ))

answered Nov 29 '19 at 09:26

Tony Cronin

1,623
1
24
30

Applying dplyr's rename to all columns while using pipe operator

Desired result

5 Answers5

Dplyr

Janitor

Linked

Related