replace_na not working

I have an R dataframe. Several columns have binary values (e.g. ‘Y’ or ‘N’). Some fields in these binary columns had NULLs (NA) values. I wanted to change the NULLs to ‘N’.

I thought the task was obvious: just use replace_na, right?

No. I tried replace_na and is.na and mutate...coalesce.... Nothing took. I didn’t get errors, but the NULLs remained. I tried banging my head on the desk, but that didn’t help, either.

Finally, while testing with a small, dummy test set, I finally got some errors and, therefore, a clue to the problem.

Long story short, I needed to convert the columns from factor to character datatypes before trying to replace the NULLs. After converting the datatypes, NULL replacement worked just fine.

Try it yourself …

Create test

# create test data.frame 
tdf <- data.frame(col1=letters[1:3], col2=c(NA, "Y", NA))
# view data.frame
tdf

 col1   col2
<fctr> <fctr>
a      NA
b      Y
c      NA

3 rows

Problem

# replace NA's with 'N'
tdf$col2 <- tdf$col2 %>% replace_na('N')

invalid factor level, NA generated

Solution

# convert column to character before replace_na, then back to factor
tdf$col2 <- as.factor(as.character(tdf$col2) %>% replace_na('N'))
# display results
tdf

 col1   col2
<fctr> <fctr>
a      N
b      Y
c      N

3 rows