efficient code to merge levels of a factor variable to form a factor variable with fewer levels.

[hasibul212003]

I have a factor variable called disease severity with 6 levels: asymptotic, healthy control,mild,moderate, severe, critical severe.
I want to form a new factor variable with four levels: asymptotic,combine(mild & moderate), severe, critical severe & ignore healthy control.

I can do this with an if else if statement but is there an easier way(efficient code) to do this?

[elmstedt]

You can just set the factor levels you aren't interested in to NA

See the example below as a guide.

set.seed(123)
x <- factor(sample(letters[1:6], 20, TRUE))
x
#>  [1] c f c b b f c e d f f a b c e c c a d a
#> Levels: a b c d e f
levels(x)[c(3, 5)] <- NA
x
#>  [1] <NA> f    <NA> b    b    f    <NA> <NA> d    f    f    a    b    <NA> <NA>
#> [16] <NA> <NA> a    d    a   
#> Levels: a b d f

^{Created on 2020-09-01 by the reprex package (v0.3.0)}

[hasibul212003]

Thanks. Is there a way to merge two levels of a factor to a single level? Say merge moderate & mild to one new level called medium.

[elmstedt]

set.seed(123)
x <- factor(sample(letters[1:6], 20, TRUE))
x
#>  [1] c f c b b f c e d f f a b c e c c a d a
#> Levels: a b c d e f
levels(x)[c(3, 5)] <- levels(x)[c(2, 4)]
x
#>  [1] b f b b b f b d d f f a b b d b b a d a
#> Levels: a b d f

^{Created on 2020-09-01 by the reprex package (v0.3.0)}

[siddharthprabhu]

The tidyverse package forcats has a number of useful functions for working with factors. For combining two levels, we could use fct_collapse().

library(dplyr, warn.conflicts = FALSE)
#> Warning: package 'dplyr' was built under R version 3.6.3
library(forcats)
#> Warning: package 'forcats' was built under R version 3.6.3

df <- tibble(disease_severity = factor(c("asymptotic", "healthy control", "healthy control", 
                                         "mild", "moderate", "severe", "critical severe", 
                                         "severe", "mild", "asymptotic", "moderate")))

glimpse(df$disease_severity)
#>  Factor w/ 6 levels "asymptotic","critical severe",..: 1 3 3 4 5 6 2 6 4 1 ...

df$disease_severity <- fct_collapse(df$disease_severity, medium = c("mild", "moderate"))
  
glimpse(df$disease_severity)
#>  Factor w/ 5 levels "asymptotic","critical severe",..: 1 3 3 4 4 5 2 5 4 1 ...

^{Created on 2020-09-02 by the reprex package (v0.3.0)}

[nirgrahamuk]

library(forcats) allows

set.seed(123)
x <- factor(sample(letters[1:6], 20, TRUE))
x

forcats::fct_collapse(x,
                      c_d = c("c","d"))
# [1] c_d f   c_d b   b   f   c_d e   c_d f   f   a   b   c_d e   c_d c_d a   c_d a  
# Levels: a b c_d e f