Compute distance between rows of a matrix.

This function was lifted from the package factoextra and slightly adjusted. The default method was changed to "spearman" and an option was added to replace missing values in the resulting distance matrix with an arbitrary value. This option defaults to 0 so that the function can be called by iheatmapr::add_row_clustering and iheatmapr::add_col_clustering.

Usage

compute_distances(
  x,
  method = "spearman",
  use = "pairwise.complete.obs",
  stand = FALSE,
  dummy = 0,
  ...
)

Arguments

x

numeric matrix

method

character string specifying the distance measure to be used; must be on of: "euclidean", "maximum", "manhattan", "canberra", "binary", "minkowski", "pearson", "spearman" or "kendall"

use

character string specifying the method for computing covariances in the presence of missing values. Only used when "pearson", "spearman" or "kendall" chosen as a distance measure

stand

logical flag specifying whether the data should be standardized; if TRUE, columns are converted to z scores using scale

dummy

value to substitute for missing values; defaults to 0

...

arguments passed to internally called functions

Value

Object of class dist. NA and NaN are substituted by the value of dummy.

Control

The function offers different distance measures but since it is called internally, it will always use default arguments. As a result, should you want to do anything other than compute Spearman correlation distance and replace NAs with 0, you must edit this function definition accordingly. At the moment (December 2020) this is sufficient. If the control were to be given to the user, remove defaults for this definition and create a wrapper within plotly_metric_clustering.

The function was originally named computeDistances

See also

[factoextra::get_dist()]

Author

Alboukadel Kassambara alboukadel.kassambara@gmail.com

Examples

x <- matrix(1:9, nrow = 3, ncol = 3)
rownames(x) <- letters[seq(NROW(x))]
compute_distances(x)
#>   a b
#> b 0  
#> c 0 0