microbenchmarking various ways of creating columns depending on values in another column of a data.table

microbenchmarking_ifelse-if_else-fifelse_fcase_match_and_chmatch.r

library(microbenchmark)
nrows <- 10^5
tst1 <- data.table::data.table(
  char = c("bb", "ds", "ok", "pb")[sample(1:4, nrows, replace = TRUE)]
)
tst2 <- tst3 <- tst4 <- tst5 <- tst6 <- data.table::copy(tst1)

microbenchmark(
   times = 200L,
  "base::ifelse" = tst1[, new := base::ifelse(char == "bb", "level1",
    base::ifelse(char == "ds", "level2",
      base::ifelse(char == "ok", "level3", "level4")
    )
  )],
  "dplyr::if_else" = tst3[, new := dplyr::if_else(char == "bb", "level1",
    dplyr::if_else(char == "ds", "level2",
      dplyr::if_else(char == "ok", "level3", "level4")
    )
  )],
  "data.table::fifelse" = tst2[, new := data.table::fifelse(char == "bb", "level1",
    data.table::fifelse(char == "ds", "level2",
      data.table::fifelse(char == "ok", "level3", "level4")
    )
  )],
  "data.table::fcase" = tst4[, new := data.table::fcase(
     char == "bb", "level1",
     char == "ds", "level2",
     char == "ok", "level3",
     default = "level4"
     )],
  "base::match" = tst5[, new := c("level1", "level2", "level3", "level4")[base::match(char, c("bb", "ds", "ok", "pb"))]],
  "data.table::chmatch" = tst6[, new := c("level1", "level2", "level3", "level4")[data.table::chmatch(char, c("bb", "ds", "ok", "pb"))]]
)

# base::match is the fastest for small (10^2) and large (10^7) columns. data.table::fcase is the most readable and allows using expressions/functions which match does not(?depends on what is put in the vector of values).

Unit: milliseconds

expr	min	lq	mean	median	uq	max	neval
base::ifelse	85.2767	98.56945	114.245198	106.44735	119.03000	285.4805	200
dplyr::if_else	18.0696	25.30790	33.638454	29.59110	33.40365	120.1349	200
data.table::fifelse	5.3107	6.37775	8.843653	7.60870	9.66705	39.8622	200
data.table::fcase	4.4215	5.64410	7.818536	6.60635	8.13225	27.8380	200
base::match	1.7898	2.22105	3.871345	2.65265	3.79105	42.6920	200
data.table::chmatch	1.9193	2.22685	3.839901	2.80290	3.56775	92.9833	200