Is there an R dplyr method for merge with all=TRUE?

Question

I have two R dataframes I want to merge. In straight R you can do:

cost <- data.frame(farm=c('farm A', 'office'), cost=c(10, 100))
trees <- data.frame(farm=c('farm A', 'farm B'), trees=c(20,30))
merge(cost, trees, all=TRUE)

which produces:

    farm cost trees
1 farm A   10    20
2 office  100    NA
3 farm B   NA    30

I am using dplyr, and would prefer a solution such as:

left_join(cost, trees)

which produces something close to what I want:

    farm cost trees
1 farm A   10    20
2 office  100    NA

In dplyr I can see left_join, inner_join, semi_join and anti-join, but none of these does what merge with all=TRUE does.

Also - is there a quick way to set the NAs to 0? My efforts so far using x$trees[is.na(x$trees)] <- 0; are laborious (I need a command per column) and don't always seem to work.

thanks

Answer 1

The most recent version of dplyr (0.4.0) now has a full_join option, which is what I believe you want.

cost <- data.frame(farm=c('farm A', 'office'), cost=c(10, 100))
trees <- data.frame(farm=c('farm A', 'farm B'), trees=c(20,30))
merge(cost, trees, all=TRUE)

Returns

> merge(cost, trees, all=TRUE)
        farm cost trees
    1 farm A   10    20
    2 office  100    NA
    3 farm B   NA    30

And

library(dplyr)
full_join(cost, trees)

Returns

> full_join(cost, trees)
Joining by: "farm"
    farm cost trees
1 farm A   10    20
2 office  100    NA
3 farm B   NA    30
Warning message:
joining factors with different levels, coercing to character vector

Answer 2

library(plyr)
> dat <- join(cost, trees, type = "full")
Joining by: farm
> dat
    farm cost trees
1 farm A   10    20
2 office  100    NA
3 farm B   NA    30

> dat[is.na(dat)] <- 0
> dat
    farm cost trees
1 farm A   10    20
2 office  100     0
3 farm B    0    30