Apply a function via hash table

Extracts the unique values of x, applies fn to those unique values, and maps the results back to the original vector. This avoids redundant computation when x contains many repeated values.

Usage

apply_fn_via_hash_table(x, fn, ...)

Arguments

x

A vector whose values need a function applied.

fn

A vectorised function, called once on the vector of unique values of x.

...

Additional arguments passed to fn.

Value

A vector of the same length as x, holding the value of fn for each element (computed via unique-value lookup). The result takes its class from the value fn returns.

Details

The function calls fn exactly once, on the vector of unique values of x. It does not call fn once per element. It does not call fn once per unique value either. fn MUST therefore be vectorised. It receives the whole vector of unique values in a single call.

This can dramatically speed up computation when there is a lot of data and a limited number of unique values. The work fn does scales with the number of unique values, not with the length of x.

See also

vignette("csutil", package = "csutil"), which times this function against a direct call to format().

Examples

x <- c("a", "b", "a", "c", "b", "a")
apply_fn_via_hash_table(x, toupper)
#> [1] "A" "B" "A" "C" "B" "A"

# passing extra arguments to fn
nums <- c(1.1, 2.2, 1.1, 3.3)
apply_fn_via_hash_table(nums, round, digits = 0)
#> [1] 1 2 1 3