microbenchmarking data.table::uniqueN(x) vs. length(unique(x))

microbenchmarking_uniqueN.r

library(microbenchmark)
nrows <- 10^5
tst <- data.table::data.table(
   char = c("bb", "ds", "ok", "pb")[sample(1:4, nrows, replace = TRUE)]
)

microbenchmark(times = 1000L,
               "base" = tst[, length(unique(char))],
               "data.table" = tst[, data.table::uniqueN(char)]
)
# data.table alternative faster (+-2) only above 10^6 rows

Unit: milliseconds

expr	min	lq	mean	median	uq	max	neval
base	1.3939	2.0471	2.904923	2.42835	2.99170	21.9820	1000
data.table	1.5846	2.5092	2.936799	2.87310	3.15655	16.8523	1000