Suppose I have the following database:
df <- data.frame(categorias=c("A","B","C","D","E"),
valores=seq(1:5))
When I make a subset of that date frame the categories I've removed continue.
subdf <- subset(df, valores <= 3)
levels(subdf$categorias)
[1] "A" "B" "C" "D" "E"
You can use the droplevels
subdf <- droplevels(subset(df, valores <= 3))
Result:
levels(subdf$categorias)
[1] "A" "B" "C"
The advantage is that this works for more than one factor variable at the same time. For example, if your data.frame is:
df <- data.frame(categorias=c("A","B","C","D","E"),
categorias2 = c("F", "G", "H", "I", "J"),
valores=seq(1:5),
valores2=rnorm(5))
If you only do subset both categorias and categorias2 would get more levels . With subdf <- droplevels(subset(df, valores <= 3)) this is resolved for all columns of factor .