Question

I have the following code:

set.seed(47)
df <- data.frame(V1 = sample(letters[1:5], size = 10, replace = TRUE),
                 V2 = 101:110)

partial_key <- data.frame(V1 = letters[1:3], V2 = 1:3)

> df
   V1  V2
1   e 101
2   b 102
3   d 103
4   e 104
5   c 105
6   d 106
7   b 107
8   c 108
9   c 109
10  e 110


> partial_key
  V1 V2
1  a  1
2  b  2
3  c  3

I’d like to replace the values of V2 in df with the corresponding values from partial_key that match in the V1 columns. The non-matches should remain as is.

With a complete key, I’d use match, which replaces the correct values, but replaces the non-matches with NA.

df[, "V2"] <- partial_key[match(df$V1, partial_key$V1), "V2"]
## Replaces too much

I can hack together a solution with %in%, but is there a better way? Something more intuitive, with less typing?

df[df$V1 %in% partial_key$V1, "V2"] <-
partial_key[match(df$V1[df$V1 %in% partial_key$V1], partial_key$V1), "V2"]
## Works, but is there a better way?
> df
   V1  V2
1   e 101
2   b   2
3   d 103
4   e 104
5   c   3
6   d 106
7   b   2
8   c   3
9   c   3
10  e 110

Answer 1

Using %in% is unnecessary since the output of match already contains that info. So you can do something like this:

replacement <- partial_key$V2[match(df$V1, partial_key$V1)]
df$V2 <- ifelse(is.na(replacement), df$V2, replacement)

Sometimes I wish R had a base if.na function similar to Excel’s IFERROR. I have it in my Rprofile:

if.na <- function(value, value.if.na) ifelse(is.na(value), value.if.na, value)
df$V2 <- if.na(replacement, df$V2)