CODEWITHSUNDEEP
rdplyr
user2464290
user2464290

Reputation: 13

How to replace lookup based on row names when using dplyr

It has been stated several times that dplyr will drop rownames, and now version 0.3 has done so.

I have often used row names to translate between different identifiers kept in a data frame like this:

test <- data.frame(Greek = c("Alpha", "Beta", "Gamma"), Letters = LETTERS[1:3])
rownames(test) <-test$Letters
lookup <- c("C", "B")
test[lookup, "Greek"]
[1] Gamma Beta 
Levels: Alpha Beta Gamma

Due to a lack of rownames this now fails with dplyr

library(dplyr)
test <- tbl_df(data.frame(Greek = c("Alpha", "Beta", "Gamma"), Letters = LETTERS[1:3]))
rownames(test) <-test$Letters
lookup <- c("C", "B")
test[lookup, "Greek"]
Source: local data frame [2 x 1]

 Greek
1    NA
2    NA

I've tried using filter() and select(), but couldn't find a solution which preserves the order of the lookup.

Upvotes: 1

Views: 2463

Answers (1)

James
James

Reputation: 66874

This is one time where you can play with matches:

test[match(lookup,test$Letters),"Greek"]
[1] Gamma Beta 
Levels: Alpha Beta Gamma

And you can wrap in a do to make it dplyr-ic:

test %>% do(`[`(.,match(lookup,.$Letters),)) %>% select(Greek)
Source: local data frame [2 x 1]

  Greek
1 Gamma
2  Beta

Or as @hadley mentions, left_join does what you are looking for:

left_join(data.frame(Letters=lookup),test) %>% select(Greek)
Joining by: "Letters"
  Greek
1 Gamma
2  Beta

Upvotes: 2

Related Questions