data.table fcase versus dplyr case_when

The following code throws an error when evaluating the line with fcase.

library(data.table) library(dplyr) tbl.test <- data.table(x = 0.5*(1:2)) tbl.test[, .(fcase(x < 1, x+1, default = x))] tbl.test[, .(case_when(x < 1 ~ x+1, TRUE ~ x) )] 

As I use case_when in this way (for most complex stuff ..), I wanted to switch to fcase, hoping a real gain in performance ..

Does anyone see where the devil hide with this usage of fcase ? The point is that I can't use a column of my table as default value with fcase ..

1 Answer

You don't explicitly state a question, but I assume you'd like to know if there's a way to use a vectorized default value with fcase(). One way to do that would be to construct a vector of TRUEs of equal length to your other conditions as the last element, similarly to how case_when() works:

library(data.table) library(dplyr) set.seed(123) tbl.test <- data.table(x = rnorm(1e6)) bench::mark( fcase = tbl.test[, .(fcase(x < 1, x + 1, rep_len(TRUE, length(x)), x))], case_when = tbl.test[, .(case_when(x < 1 ~ x + 1, TRUE ~ x))] ) #> Warning: Some expressions had a GC in every iteration; so filtering is disabled. #> # A tibble: 2 x 6 #> expression min median `itr/sec` mem_alloc `gc/sec` #> <bch:expr> <bch:tm> <bch:tm> <dbl> <bch:byt> <dbl> #> 1 fcase 16.5ms 24.3ms 33.3 36.1MB 43.1 #> 2 case_when 146.4ms 147.9ms 6.60 129.9MB 28.1