Aggregate all columns of data.table, without having to reference them by name

Question

I'd like to do the equivalent of the following, but with data.table's "by":

dt <- data.table(V1=rnorm(100), V2=rnorm(100), V3=rnorm(100), ...
                 group=rbinom(100,2,.5))
dt.agg <- aggregate(dt, by=list(dt$group), FUN=mean)

I know that I could do this:

dt.agg <- dt[, list(V1=mean(V1), V2=mean(V2), V3=mean(V3)), by=group]

But for the case I'm considering I have 100 or so columns V1-V100 (and I always want to aggregate all of them by a single factor, as in aggregate above) so the data.table solution I've got above isn't feasible.

Answer 1

dt[, lapply(.SD, mean), by=group]

To specifiy columns:

dt[,...,by=group, .SDcols=c("V1", "V2", "V3", ...)]
dt[,...,by=group, .SDcols=names(dt)[1:100]]