How to join NA values in R

Question

I have two tables that are joined. After the join, some of the values come out as NA.

I am trying to join again with a third data set, but only on those NA values. How do I do it?

The joined results

library(plyr)

## first table
original_value <- c('old_a', 'old_b', 'old_c', 'old_d') 
key <- c('a', 'b', 'c', 'd')
data <- data.frame(key, original_value, stringsAsFactors = FALSE)

## lookup table
new_value <- c('new_a', 'new_b')
key <- c('a', 'b')
lookup <- data.frame(key, new_value, stringsAsFactors = FALSE)

## the joined data
data_lookup_joined <- join(data, lookup, by = "key")

> data_lookup_joined
  key original_value new_value
1   a          old_a     new_a
2   b          old_b     new_b
3   c          old_c      <NA>
4   d          old_d      <NA>

This is the output I am trying to get:

## a third data set to join the NA values
unmatched_value <- c('unmatched_c', 'unmatched_d')
key <- c('c', 'd')
unmatched_lookup <- data.frame(key, unmatched_value, stringsAsFactors = FALSE)

 key original_value new_value
1   a          old_a     new_a
2   b          old_b     new_b
3   c          old_c     unmatched_c
4   d          old_d     unmatched_d

This is what I have tried that did not work.

data_lookup_joined$new_value [is.na(data_lookup_joined$new_value)] <- join(data_lookup_joined, unmatched_lookup, by = "key")

What do I need to do?

Answer 1

# join the rows with missing values
has_na = is.na(data_lookup_joined$new_value)
na_join = join(data_lookup_joined[has_na, c("key", "original_value")], unmatched_lookup) 
# make the column names match
names(na_join)[3] = "new_value"

# put it back together
final_result = rbind(data_lookup_joined[!has_na, ], na_join)

Of course, the simpler way would be to rbind lookup and unmatched_lookup first, then you just need one join.