Merge 2 data.tables with more than one column the data.table way

Question

I have two data.table s as following:-

a <- data.table(id = 1:10, val = 2010:2019)
b <- data.table(id = c(1, 2, 4, 6), year = 1:4)

Now if I merged b and a as following:-

b[a, val := i.val, on = "id"]

This will make an extra column in b called val . This will also not reassign the memory for b data.table .

I wanted to know if a more than 2 columns and was as following:-

    a <- data.table(id = 1:10, val = 2010:2019,
                    twr = c(10, 13, 22 ,43, 23, 23, -4, 33, -54, 34))

how to merge the two data.table s ( b and a ), the data.table way, ie not using merge or any of the join function.

But using [, , on = "id"] syntax.

I want to know this because using any of the join functions or merge makes a whole new object, whereas the data.table way only creates the new columns and not a whole new object.

Thanks in advance.

Answer 1

If there are only two columns to be returned, just wrap then in a list (or short form .( ) after joining on by 'id', and assign := ) those columns to 'b'

b[a, names(a)[-1] := .(i.val, i.twr), on = .(id)]

---

If there are many columns to be returned

nm1 <- names(a)[-1]
b[a, (nm1) := mget(paste0("i.", nm1)), on = .(id)]

-ouput

b
   id year  val twr
1:  1    1 2010  10
2:  2    2 2011  13
3:  4    3 2013  43
4:  6    4 2015  23

Answer 2

With development version 1.14.1, data.table has gained the env parameter which is meant for programming on data.table :

cols <- setdiff(names(a), "id")
b[a, on = "id", (cols) := acols, env = list(acols = as.list(cols))][]

 id year val twr 1: 1 1 2010 10 2: 2 2 2011 13 3: 4 3 2013 43 4: 6 4 2015 23

This will work in many cases where there no duplicate column names in a and b except those to join on. However, we can explicitely refer to columns of a by using the prefix i. :

b[a, on = "id", (cols) := acols, env = list(acols = as.list(paste0("i.", cols)))][]