R dplyr: rename variables using string functions

Question

(Somewhat related question: Enter new column names as string in dplyr's rename function)

In the middle of a dplyr chain (%>%), I would like to replace multiple column names with functions of their old names (using tolower or gsub, etc.)

library(tidyr); library(dplyr)
data(iris)
# This is what I want to do, but I'd like to use dplyr syntax
names(iris) <- tolower( gsub("\\.", "_", names(iris) ) )
glimpse(iris, 60)
# Observations: 150
# Variables:
#   $ sepal_length (dbl) 5.1, 4.9, 4.7, 4.6, 5.0, 5.4, 4.6,...
#   $ sepal_width  (dbl) 3.5, 3.0, 3.2, 3.1, 3.6, 3.9, 3.4,...
#   $ petal_length (dbl) 1.4, 1.4, 1.3, 1.5, 1.4, 1.7, 1.4,...
#   $ petal_width  (dbl) 0.2, 0.2, 0.2, 0.2, 0.2, 0.4, 0.3,...
#   $ species      (fctr) setosa, setosa, setosa, setosa, s...

# the rest of the chain:
iris %>% gather(measurement, value, -species) %>%
  group_by(species,measurement) %>%
  summarise(avg_value = mean(value)) 

I see ?rename takes the argument replace as a named character vector, with new names as values, and old names as names.

So I tried:

iris %>% rename(replace=c(names(iris)=tolower( gsub("\\.", "_", names(iris) ) )  ))

but this (a) returns Error: unexpected '=' in iris %>% ... and (b) requires referencing by name the data frame from the previous operation in the chain, which in my real use case I couldn't do.

iris %>% 
  rename(replace=c(    )) %>% # ideally the fix would go here
  gather(measurement, value, -species) %>%
  group_by(species,measurement) %>%
  summarise(avg_value = mean(value)) # I realize I could mutate down here 
                                     #  instead, once the column names turn into values, 
                                     #  but that's not the point
# ---- Desired output looks like: -------
# Source: local data frame [12 x 3]
# Groups: species
# 
#       species  measurement avg_value
# 1      setosa sepal_length     5.006
# 2      setosa  sepal_width     3.428
# 3      setosa petal_length     1.462
# 4      setosa  petal_width     0.246
# 5  versicolor sepal_length     5.936
# 6  versicolor  sepal_width     2.770
# ... etc ....  

Answer 1

This is a very late answer, on May 2017

As of dplyr 0.5.0.9004, soon to be 0.6.0, many new ways of renaming columns, compliant with the maggritr pipe operator %>%, have been added to the package.

Those functions are:

• rename_all
• rename_if
• rename_at

There are many different ways of using those functions, but the one relevant to your problem, using the stringr package is the following:

df <- df %>%   rename_all(       funs(         stringr::str_to_lower(.) %>%         stringr::str_replace_all(., '\\.', '_')       )   ) 

And so, carry on with the plumbing :) (no pun intended).

Answer 2

I think you're looking at the documentation for plyr::rename, not dplyr::rename. You would do something like this with dplyr::rename:

iris %>% rename_(.dots=setNames(names(.), tolower(gsub("\\.", "_", names(.))))) 

Answer 3

Here's a way around the somewhat awkward rename syntax:

myris <- iris %>% setNames(tolower(gsub("\\.","_",names(.))))

Answer 4

As of 2020, rename_if, rename_at and rename_all are marked superseded. The up-to-date way to tackle this the dplyr way would be rename_with():

iris %>% rename_with(tolower)

or a more complex version:

iris %>% 
  rename_with(stringr::str_replace, 
              pattern = "Length", replacement = "len", 
              matches("Length"))

(edit 2021-09-08)
As mentioned in a comment by @a_leemo, this notation is not mentioned in the manual verbatim. Rather, one would deduce the following from the manual:

iris %>% 
  rename_with(~ stringr::str_replace(.x, 
                                     pattern = "Length", 
                                     replacement = "len"), 
              matches("Length")) 

Both do the same thing, yet, I find the first solution a bit more readable. In the first example pattern = ... and replacement = ... are forwarded to the function as part of the ... dots implementation. For more details see ?rename_with and ?dots.