Compute Gower Distance

Computes the Gower (1971) distance matrix for mixed-type datasets. Supports missing values (NA) by setting their weight contribution to zero.

Usage

zd_gower(x, y = NULL, weight = NULL, num_threads = 1L)

Arguments

x

A data frame or matrix of size N_x x P.

y

An optional data frame or matrix of size N_y x P.

weight

An optional numeric vector of weights for each column. If NULL, all columns are weighted equally.

num_threads

An integer specifying the number of threads to use. Defaults to 1L (single-threaded).

Value

A numeric matrix of Gower distances. If y is NULL, returns an N_x x N_x matrix. If y is provided, returns an N_x x N_y matrix.

References

Gower, J. C. (1971). A general coefficient of similarity and some of its properties. Biometrics, 27(4), 857–871. doi:10.2307/2528823

Podani, J. (1999). Extending Gower's general coefficient of similarity to ordinal characters. Taxon, 48(2), 331–340. doi:10.2307/1224024

Examples

df_x <- data.frame(
  a = c(1, 2, 5),
  b = factor(c("A", "B", "A")),
  c = c(TRUE, FALSE, TRUE)
)
zd_gower(df_x, num_threads = 1L)
#>           [,1]      [,2]      [,3]
#> [1,] 0.0000000 0.7500000 0.3333333
#> [2,] 0.7500000 0.0000000 0.9166667
#> [3,] 0.3333333 0.9166667 0.0000000

df_y <- data.frame(
  a = c(2, 4),
  b = factor(c("A", "C")),
  c = c(FALSE, TRUE)
)
zd_gower(df_x, df_y, weight = c(2.0, 0.5, 0.5), num_threads = 2L)
#>           [,1]      [,2]
#> [1,] 0.3333333 0.6666667
#> [2,] 0.1666667 0.6666667
#> [3,] 0.6666667 0.3333333