good.dist: A distance function that, unlike the built-in dist: 1)...

In NathanWycoff/mds.methods:

Description

A distance function that, unlike the built-in dist: 1) Returns a matrix instead of a dist object (I need the diagonals for inner prod distance) 2) Allows for custom distance function specification.

Usage

good.dist(X, dist.func, weights = NULL, symm = FALSE)

Arguments

X	An n by p real valued matrix, rows representing observations, columns representing features. A distance matrix will be computed along its rows.
dist.func	A function which accepts two arguments: two rows X, and returns a distance (or generalized distance) between them.
weights	Either a length p nonnegative vector or NULL. If not null, the ith column of X will be scaled by the root of the ith element of weights prior to distance calculation. If not, no scaling will occur.
symm	Is the distance function symmetric? If so, we can skip a bunch of function evaluations. Default is FALSE.

Value

An n by n matrix of distances, element i,j of which being the result of dist.func applied to rows i and j.

NathanWycoff/mds.methods documentation built on May 23, 2019, 7:32 a.m.