Apply multiple functions to multiple columns (with data.table)

mult-fxn-mult-col.R

my.summary = function(x) list(mean = mean(x), median = median(x))
DT[, as.list(unlist(lapply(.SD, my.summary))), .SDcols = c('a', 'b')]

N.B.: This can be slow for very large datasets – the as.list(unlist()) formulation is the culprit. Another formulation, which is a bit more code-heavy but better in terms of performance, is to melt the relevant columns and compute on the melted data.table.