I have a dataframe in a certain order:
df <-
data.frame(
foo = 1:3,
bar = LETTERS[1:3],
baz = rnorm(3)
)
df
foo bar baz
1 1 A 0.41474174
2 2 B -0.08416768
3 3 C -0.27931232
In another dataframe, I have the old variable names matched to some new names, but in a different order:
variable_match <-
data.frame(
old = names(df)[c(2, 3, 1)],
new = LETTERS[1:3]
)
variable_match
old new
1 bar A
2 baz B
3 foo C
My question is: How do I rename the variables in the original dataframe by looking up the corresponding value in the second dataframe. I'm ideally looking for a tidyverse solution. I have tried variations of:
library(tidyverse)
df %>% rename_at(variable_match$old, funs(variable_match$new))
assuming that rename_at would be the right approach, but this doesn't work. I'm wondering if purrr::map_* would be the right approach, but can't see how. Many thanks for your suggestions.
Here is a one-line base solution:
names(df2) = variable_match$new[match(names(df), variable_match$old)]
It may not be "ideal" for you (it doesn't need the tidyverse to work), but it is simple and doesn't require loading any extra packages, instead relying on common built-in functions.
As noted in comments, if you prefer a nested statement with pipes (aren't pipes intended to improve readability and prevent nesting?) the simple line above is equivalent to
library(purrr)
library(dplyr)
library(magrittr)
df = df %>%
set_names(
var_match %>%
pull(new) %>%
extract(
names(df) %>%
match(var_match$old)
)
)
I'm a big fan of pipes and dplyr - I use them consistently when the make things simpler and more readable. In this case they take a simple line and turn it into a programming puzzle, both in how to write it and how to read it.
A nicer interface overall is the data.table::setnames function. If you convert to a datatable, then the code is setnames(df, old = var_match$old, new = var_match$new). This is robust in case not all names are changed (see comments below).