Switch in

About Switch

The switch() function can be a concise replacement for a long series of if ... else if tests. The variable being switched on is most commonly a string, and if so the quotes can be omitted from the selector.

star <- function(type) {
  switch(type,
    M = , # will "fall through" if no value given
    K = "red dwarf",
    G = "Earth-like",
    "bigger star" # only correct for O,B,A,F
  )
}

> star("M")
[1] "red dwarf"

Note that options will only fall through if the value is left blank, as with M in the example above. There is no need to include break statements as with some other languages.

The final value can be a default, as here, or a stop() to throw an error if the conditions are intended to be exhaustive. Switching on an integer is slightly different: for these the default is always NULL (which will be discussed in the Nothingness Concept).

Functions in dplyr

The switch in base R is quite limited: it will only do an exact match to a single input. This seemed reasonable when R was first released in 1993, but needs improvement for modern usage.

The dplyr library provides two extra functions related to switch.

The recode_values and replace_values functions

These two related functions allow a vectorized switch-like mapping of old values to new values.

The main difference between them is that recode_values() creates an entirely new vector, while replace_values() allows partial updates of an existing vector.

Matching is still exact, but:

• The options on the left can be vectors, and the input matches if any element matches.
• Options on the left and results on the right are linked with a tilde ~ instead of =.
• Missing values NA can be matched explicitly (this will be discussed in the Nothingness Concept).
• A default can be specified in all cases.

library(dplyr)

x <- c("a", "b", "a", "d", "b", NA, "c", "e", "z")
recode_values(
  x,
  "a" ~ 1,
  "b" ~ 2,
  "c" ~ 3,
  c("d", "e") ~ 4, # either "d' or "e" will match
  NA ~ 0,          # matches missing values
  default = 100   # note the different syntax for the default
)
#> [1]   1   2   1   4   2   0   3   4 100

The case_when function

case_when takes case_match syntax a stage further, by allowing any logical expression on the left of the ~.

The input vector (x in the example below) is not supplied as an argument, it just needs to be already defined.

x <- 1:10
case_when(
  x < 3 ~ "low",
  between(x, 3, 5) ~ "mid",
  between(x, 6, 8) ~ "high",
  .default = "what?"
)
#>  [1] "low"   "low"   "mid"   "mid"   "mid"   "high"  "high"  "high"  "what?" "what?"

The between() function is also part of dplyr.

String functions were already discussed and Regular Expressions will be the subject of a later concept. These combine very well with case_when.

library(stringr)

tracks <- "R C Python C# Elixir C++ Odin" |> str_split_1(" ")
tracks
#> [1] "R"      "C"      "Python" "C#"     "Elixir" "C++"    "Odin"  

case_when(
  str_starts(tracks, "C") ~ "C-like",
  str_starts(tracks, "[AEIOU]") ~ "vowel",
  .default = "other"
)
#> [1] "other"  "C-like" "other"  "C-like" "vowel"  "C-like" "vowel"