Question

I have the following data.table in R:

  library(data.table)
  DT = data.table(x=rep(c("b","a","c"),each=3), y=sample(rnorm(9)), v=1:9)

I just want to compute the minimum and the maximum by the column x and add these two new columns to DT. Here is my line for this:

  DT[,c("e","d"):= list(min(y),max(y)), with=FALSE, by = x] 
  Error in `[.data.table`(DT, , `:=`(c("e", "d"), list(min(y), max(y))),  : 
  'with' must be TRUE when 'by' or 'keyby' is provided

Nonetheless, if I write: DT[,c("e","d"):= list(min(y),max(y)), with=FALSE], I get this:

     x       y    v     e       d
  1: a -1.7125000 4 -1.7125 1.30553
  2: a  1.0198038 5 -1.7125 1.30553
  3: a  1.3055301 6 -1.7125 1.30553
  4: b -0.9238759 1 -1.7125 1.30553
  5: b  0.3077016 2 -1.7125 1.30553
  6: b -1.2580845 3 -1.7125 1.30553
  7: c -0.9399120 7 -1.7125 1.30553
  8: c -0.1910583 8 -1.7125 1.30553
  9: c  0.1239158 9 -1.7125 1.30553

As you can see, this is working but it’s not doing the task by x. I want to obtain something similar but e and d are supposed to be computed by each value of the variable x. So, my question is: How can I solve this?

Answer 1

“:= by group” (new in version 1.8.2) and “:= with multiple new columns” (new in version 1.7.8) are both relatively recent additions to data.table.

“:= by group with multiple new columns” just hasn’t (yet) been implemented.

So for now, you can either do this (if you want a one-liner):

setkey(DT, "x")
DT <- DT[DT[,list(e=min(y), d=max(y)), by=key(DT)]]

or this (if you want to minimize extra copying operations):

setkey(DT, "x")
DT[,e:=min(y), by=key(DT)]
DT[,d:=max(y), by=key(DT)]