将 NA 转换为因子水平

Question

I have a vector with NA values that I would like to replace by a new factor level NA .

a = as.factor(as.character(c(1, 1, 2, 2, 3, NA)))
a
[1] 1    1    2    2    3    <NA>
Levels: 1 2 3

This works, but it seems like a strange way to do it.

a = as.factor(ifelse(is.na(a), "NA", a))
class(a)
[1] "factor"

This is the expected output:

a
[1] 1  1  2  2  3  NA
Levels: 1 2 3 NA

Answer 1

You can use addNA() .

x <- c(1, 1, 2, 2, 3, NA)
addNA(x)
# [1] 1    1    2    2    3    <NA>
# Levels: 1 2 3 <NA>

This is basically a convenience function for factoring with exclude = NULL . From help(factor) -

addNA modifies a factor by turning NA into an extra level (so that NA values are counted in tables, for instance).

So another reason this is nice is because if you already have a factor f , you can use addNA() to quickly add NA as a factor level without changing f . As mentioned in the documentation, this is handy for tables. It also reads nicely.

Answer 2

You can add the NA as a level and change the level name to something more explicit than <NA> using fct_explicit_na from package forcats .

library(forcats)

By default you get the new level as (Missing) :

fct_explicit_na(a)

[1] 1         1         2         2         3         (Missing)
Levels: 1 2 3 (Missing)

You can set it to something else:

fct_explicit_na(a, "unknown")

[1] 1       1       2       2       3       unknown
Levels: 1 2 3 unknown

Answer 3

Set the exclude argument to NULL to include NAs as levels (and use factor instead of as.factor. Does the same thing and has more arguments to set):

a = factor(as.character(c(1, 1, 2, 2, 3, NA)), exclude = NULL)

> a
[1] 1    1    2    2    3    <NA>
Levels: 1 2 3 <NA>