R: Update column based on matching rows from another data frame

Question

I have

mydf1 <- data.frame(ID = c(1,2,3,4,5), color = c("red", NA, NA, NA, "green"), name = c("tom", "dick", "harry", "steve", "mike"))
mydf2 <- data.frame(ID = c(1,2,99), color = c("red", "orange", "yellow"), name = c("tom", "dick", "Aaron"))

I would like to update mydf1$color with the corresponding color from mydf2 for any rows that match on both ID and name. The desired output would be to update the color in row 2 to orange and leave the rest as is:

   ID  color   name
1  1   red     tom
2  2   orange  dick
3  3   <NA>    harry
4  4   <NA>    steve
5  5   green   mike

I tried solutions with asymmetric merging as in some previous posts, but obtained undesired overwriting of some of my fields in mydf1. I then tried using match as suggested in another post but received an error. Not sure why the match condition is not working.

mydf1$color <- mydf2$color[match(mydf1[c("ID", "name")], mydf2[c("ID", "name")])]

Answer 1

We can use a join with data.table on the 'ID' and 'name' column and update the corresponding value of 'color' from the second dataset in the first dataset by assignment ( := )

library(data.table)
setDT(mydf1)[mydf2, color := i.color, on = .(ID, name)]
mydf1
#   ID  color  name
#1:  1    red   tom
#2:  2 orange  dick
#3:  3   <NA> harry
#4:  4   <NA> steve
#5:  5  green  mike

---

match works on vector/matrix and not on data.frame . If we need to use match , then paste the 'ID', 'name' from each datasets and do a match

i1 <- match(paste(mydf1$ID, mydf1$name), paste(mydf2$ID, mydf2$name), nomatch = 0)

---

Or using tidyverse

library(dplyr)
mydf1 %>%
   left_join(mydf2, by = c("ID", "name")) %>%
    transmute(ID, name, color = coalesce(as.character(color.x), 
                                         as.character(color.y)))
#  ID  name  color
#1  1   tom    red
#2  2  dick orange
#3  3 harry   <NA>
#4  4 steve   <NA>
#5  5  mike  green

Answer 2

As of dplyr version 1.0, you can use the experimental rows_* functions. Unfortunately rows_update doesn't allow for a "inner" or "left" update--it will throw an error if there are non-matching rows. We can work around this using rows_upsert followed by a inner_join :

rows_upsert(mydf1, mydf2, by = c("ID", "name")) %>%
  inner_join(mydf1 %>% select(ID, name))
# Joining, by = c("ID", "name")
#   ID  color  name
# 1  1    red   tom
# 2  2 orange  dick
# 3  3   <NA> harry
# 4  4   <NA> steve
# 5  5  green  mike