Replace value from updated dataset based on number of instances it appears in a second dataset - r

Question

I have a simple 2-column dataset containing variable cluster_size and index. Originally all values of index were assigned a value 1. Subsequently, I received a second dataset containing only a few clusters where index should updated with different integer values.
I simply need to replace the index value from the updated dataset. My specific issue is that the values for cluster_size can repeat multiple times, but I only need to replace it for the number of instances it appears in the updated dataset. For instance, in the example data below, the cluster_size value of 34 appears three times, but only once in the updated data with an index of 6. This means that only one of these three rows should update to 6 (doesn't matter which one).
Code to recreate a 20-row sample of the original data (have), updated subset (updated), and desired dataset (want) are below. The actual data has tens of thousands of rows. Ive tried several merge and loop functions (all too pathetic to waste your time by posting here), but cant seem to find an elegant solution.
# Data with original index cases
set.seed(03151813)
have <- data.frame(clust_size=sample(1:50,20,replace=TRUE),index=rep(1,times=20))
have <- have[order(have$clust_size),]
# Updated data only contains clusters that need updating of inde
updated <- data.frame(clust_size=c(30,34,42,44,44,46),
index=c(2,6,4,8,9,4))
# Desired dataset
want <- data.frame(clust_size=have$clust_size,
index=c(rep(1,times=9),2,1,6,
1,1,1,4,1,8,9,4))

Answer 1

Here is a base R approach. Add row numbers to have and updated for each clust_size. So the clust_size of 34 will have rows numbered consecutively 1, 2, and 3.
Then, you can merge the two together on both clust_size and row number. If you include all.x you will get all rows from the first data frame have.
Final step is to replace the missing NA values in your new index column with the original index.
have$rn <- with(have, ave(seq_along(clust_size), clust_size, FUN = seq_along))
updated$rn <- with(updated, ave(seq_along(clust_size), clust_size, FUN = seq_along))
want <- merge(have, updated, all.x = TRUE, by = c("clust_size", "rn"))
want$index.y <- ifelse(is.na(want$index.y), want$index.x, want$index.y)
want[, c("clust_size", "index.y")]
An alternate version using dplyr would be something like this:
library(dplyr)
have2 <- have %>%
group_by(clust_size) %>%
mutate(rn = row_number())
updated2 <- updated %>%
group_by(clust_size) %>%
mutate(rn = row_number())
left_join(have2, updated2, by = c("clust_size", "rn")) %>%
mutate(index.y = coalesce(index.y, index.x))
Output
clust_size index.y
1 1 1
2 5 1
3 8 1
4 10 1
5 16 1
6 20 1
7 22 1
8 27 1
9 29 1
10 30 2
11 30 1
12 34 6
13 34 1
14 34 1
15 35 1
16 42 4
17 43 1
18 44 8
19 44 9
20 46 4