How to select columns programmatically in a data.table?

Question

I have the following data.table (DT):

DT <- data.table(V1 = 1:3, V2 = 4:6, V3 = 7:9)

I would like to select a subset of the variables programmatically (dynamically), by using an object where the relevant variable names are stored. For example, I want to select the two columns "V1" and "V3" stored in a variable "keep"

keep <- c("V1", "V3")

If we were to select the "keep" columns from a data.frame, the following would work:

DT[keep]

Unfortunately, this is not working when this is a data.table. I thought the data.frame and data.table are identical with this kind of behavior, but apperently they aren't. Anybody able to advise on the correct syntax?

Answer 1

This is covered in FAQ 1.1, 1.2 and 2.17.

Some possibilities:

DT[, keep, with = FALSE]
DT[, c('V1', 'V3'), with = FALSE]
DT[, c(1, 3), with = FALSE]
DT[, list(V1, V3)]

The reason DF[c('V1','V3')] works as it does for a data.frame is covered in ?`[.data.frame`

Data frames can be indexed in several modes. When [ and [[ are used with a single vector index (x[i] or x[[i]]), they index the data frame as if it were a list. In this usage a drop argument is ignored, with a warning.

---

From data.table 1.10.2, you may use the .. prefix when subsetting columns programmatically:

When j is a symbol prefixed with .. it will be looked up in calling scope and its value taken to be column names or numbers [...] It is experimental.

Thus:

DT[ , ..keep]
#    V1 V3
# 1:  1  7
# 2:  2  8
# 3:  3  9

Answer 2

Some more possibilities:

DT[, .SD, .SDcols = keep]
DT[, mget(keep)]