Rename all levels of factor variable, is there a tidyverse way to do it?

Question

I occasionally need to rename all the levels of factor variable. I know how to achieve this with R base like: levels(factor_variable) <- levels(new_variable). But I would really like to have a way to do this kind of thing using tidyverse. I look in dplyr and forcats but I do not found anything to solve it. I would like to be able to do what I achieve in example 1, but working with the %>% operator.

Example 1, with R base (which works)

my_levels <- letters
sample_data <- data.frame(factor_data = factor(sample(my_levels,size = 500,replace = T) ,
                                               levels = my_levels),
                          Any_other_data = rnorm(500))

my_new_levels <- rnorm(length(letters))

levels(sample_data$factor_data) <- levels(factor(my_new_levels))

Example 2, one thing I try and not work with tidyverse

library(tidyverse)

my_levels <- letters
sample_data <- tibble(factor_data = factor(sample(my_levels,size = 500,replace = T) ,
                                               levels = my_levels),
                          Any_other_data = rnorm(500))

my_new_levels <- rnorm(length(letters))

# Get error
sample_data <- sample_data %>%
  mutate(levels(factor_data) = levels(factor(my_new_levels)))
# Get error
sample_data <- sample_data %>%
  mutate(factor_data = recode(factor_data, levels(factor_data) = levels(factor(my_new_levels))))

I also try with recode, but it is not only manually (each value at a time), but it also does not accept the %>%operator. This is som things I try to see what happened:

sample_data <- sample_data %>%
  recode(factor_data, a = '-2.5')

sample_data <- sample_data %>%
  recode_factor(factor_data, a = '-2.5')

recode(sample_data$factor_data, levels(sample_data$factor_data) = levels(factor(my_new_levels)))

recode(sample_data$factor_data, a = '-2.5')
recode_factor(sample_data$factor_data, a = '-2.5')

The base method you show is more complicated than necessary. You really just want levels(sample_data$factor_data) <- my_new_levels. I don't see why you think a piped version would improve on this. — user2554330
For me, it would make it easier and faster to write my code and in one line make a graph with ggplot. I chase wonder if there were an easy way to do it with a pipe version. — Santiago I. Hurtado

ek-g ek-g · Accepted Answer · 2020-06-11T13:09:22

You can easily do that with a named vector and forcats::fct_recode():

library(tidyverse)
set.seed(42)

my_levels <- letters
sample_data <- data.frame(factor_data = factor(sample(my_levels,size = 500,replace = T) ,
                                               levels = my_levels),
                          Any_other_data = rnorm(500))

my_new_levels <- rnorm(length(letters))

# create a named vector with the new levels
named_level_vector <- levels(sample_data$factor_data)
names(named_level_vector) <- my_new_levels

# use mutate and fct_recode with that vector

sample_data <- sample_data %>% 
  mutate(new_factor_data = forcats::fct_recode(factor_data, !!!named_level_vector))

head(sample_data)
#>   factor_data Any_other_data    new_factor_data
#> 1           q     0.48236947  0.223521215874458
#> 2           e     0.99294364  -1.12828853519737
#> 3           a    -1.24639550  -2.55382485095083
#> 4           y    -0.03348752   1.67099730539817
#> 5           j    -0.07096218 -0.318990710826149
#> 6           d    -0.75892065  -1.17990419995829

^{Created on 2020-06-11 by the reprex package (v0.3.0)}

Rename all levels of factor variable, is there a tidyverse way to do it?

1 Answers