R data.table：重用聚合

Question

I want to apply the same aggregation to multiple data tables, without rewriting the aggregation scheme.

Consider

dt1 <- data.table(id = c(1,2), a = rnorm(10), b = rnorm(10), c = rnorm(10))
dt2 <- data.table(id = c(1,2), a = rnorm(10), b = rnorm(10), c = rnorm(10))

dt1_aggregates <- dt1[, .(mean_a=mean(a), sd_a=sd(a), mean_b=mean(b), sd_b=sd(b)), by=id]
dt2_aggregates <- dt2[, .(mean_a=mean(a), sd_a=sd(a), mean_b=mean(b), sd_b=sd(b)), by=id]

Is there some way to reuse the dt1_aggregates aggregation scheme for dt2 without having to write it out twice?

Answer 1

You can quote the expression you want, and then evaluate it within the data.table:

my.call=quote(list(mean_a=mean(a), sd_a=sd(a), mean_b=mean(b), sd_b=sd(b)))
dt1[, eval(my.call), by=id]

Produces

   id       mean_a      sd_a      mean_b      sd_b
1:  1  0.004165423 0.7504691 -0.05001424 1.4440434
2:  2 -0.430910188 0.9648096  0.26918995 0.8680997

And

dt2[, eval(my.call), by=id]

Produces

   id     mean_a     sd_a     mean_b      sd_b
1:  1  0.2974145 1.191863 -0.0588854 0.7896988
2:  2 -0.4642856 1.438937  0.3612607 1.0581702