R how to change one of the level to NA

Question

I have a data set and one of its column has factor levels \"a\" \"b\" \"c\" \"NotPerformed\". How can I change all the \"NotPerformed\" factors to

Answer 1

Or simply use the inbuilt exclude option, which works regardless of whether the initial variable is a character or factor.

x <- c("a", "b", "c", "NotPerformed")

factor(x, exclude = "NotPerformed")
[1] a    b    c    <NA>
Levels: a b c

factor(factor(x), exclude = "NotPerformed")
[1] a    b    c    <NA>
Levels: a b c

Answer 2

Set one of the levels to NA through tidyverse Pipeline, %>%.
This may serve better as a comment, but I do not have that many reputation.
In my case, the income variable is int with values of c(1:7, 9). Among the levels, "9" represents "Do not wish to answer".

## when all int should be fctr 
New_data <- data %>% mutate_if(is.integer, as.factor)  %>% 
mutate(income = fct_recode(income, NULL = "9"))

I also tried recode(), it does not work.

Answer 3

Set the level to NA:

x <- factor(c("a", "b", "c", "NotPerformed"))
x
## [1] a            b            c            NotPerformed
## Levels: a b c NotPerformed
levels(x)[levels(x)=='NotPerformed'] <- NA
x
## [1] a    b    c    <NA>
## Levels: a b c

Note that the factor level is removed.

Answer 4

I revise my old answer and provide what you can do as of September 2016. With the development of the dplyr package, now you can use recode_factor() to do the job.

x <- factor(c("a", "b", "c", "NotPerformed"))

# [1] a            b            c            NotPerformed
# Levels: a b c NotPerformed

library(dplyr)
recode_factor(x, NotPerformed = NA_character_)
# [1] a    b    c    <NA>
# Levels: a b c