recode examples

library(dplyr)
library(microbenchmark)

x = sample(c("less often", "1", "2", "3", "4", "4+"), 1e6, r = TRUE)

fun1 = function(x) { 
  case_when(
    x == "less often" ~ 1L,
    x %in% c("1", "2") ~ 2L,
    x %in% c("3", "4") ~ 3L,
    x == "4+" ~ 4L
  )
}

fun2 = function(x) { 
  y = rep(NA_integer_, length(x)) # preallocate
  y[x == "less often"] = 1L
  y[x %in% c("1", "2")] = 2L
  y[x %in% c("3", "4")] = 3L
  y[x == "4+"] = 4L
  y
}

fun3 = function(x) {
  x = factor(x, levels = c("less often", "1", "2", "3", "4", "4+"))
  levels(x) = c(1L, 2L, 2L, 3L, 3L, 4L)
  as.integer(x)
}

fun4 = function(x) {
  c(2,2,3,3,4,1)[as.numeric(as.factor(x))]
}

microbenchmark(
  fun1(x),
  fun2(x),
  fun3(x),
  fun4(x)
)

# Unit: milliseconds
#    expr       min        lq      mean     median        uq       max neval
# fun1(x) 125.72752 131.64690 153.78324 136.260120 185.57678 218.02885   100
# fun2(x)  53.44385  57.32536  62.91769  58.625194  61.54146 125.60717   100
# fun3(x)   7.57748   7.86945  11.33043   8.566502  10.97013  72.49452   100
# fun4(x)  21.72993  24.10765  28.01275  25.517953  27.13678  88.40717   100

all.equal(fun1(x), fun2(x))
all.equal(fun2(x), fun3(x))
all.equal(fun3(x), fun4(x))