Rename one level of a factor in R

Question

I'm attempting to rename the level A of factor column1 in the dataframe df in R. My current approach is this:

levels(df[!is.na(df$column1) & df$column1 == 'A',]) <- 'B'

which doesn't throw any errors or warnings but is completely ineffective.

B is not an already existing level (which from trial and error I came to suspect was important), so the following, my first attempt, didn't work either

df[!is.na(df$column1) & df$column1 == 'A', 'column1'] <- 'B'

Could anyone guide me to the correct approach?

Solution 1

I was going to suggest

levels(df$column1)[levels(df$column1)=="A"] <- "B"

or use the utility function plyr::revalue:

library("plyr")
df <- transform(df,
          column1=revalue(column1,c("A"="B")))

transform() is a little sugar that's not necessary; you could use df$column1 <- revalue(df$column1(...))

For completeness, car::recode also works, although I find it a little bit clunkier that plyr::revalue (because the recoding is specified as a quoted string).

car::recode(df$column1,"'A'='B'")

Solution 2

One way would be just to change the label of the level. First, some test data

df <- data.frame(column1=c("A","B","C","A","B"))

and now we replace "A" with "X"

levels(df$column1) <- gsub("A","X", levels(df$column1))

and we can see that it's changed

  column1
1       X
2       B
3       C
4       X
5       B

You might need to be careful with gsub() since it accepts a regular expression. A more specific replacement would be

gsub("^A$","X", levels(df$column1))

to match exactly "A" and not "CAB" or something else with a capital A.