somKmeans {kohonen}R Documentation

Refinement of codebook vectors by kmeans

Description

Do kmeans of the codebook vectors, which is equivalent to the SOM algorithm when the size of the neighbourhood is smaller than 1. Not meant to be called directly.

Usage

somKmeans(codes, data, max.iter = 20, verbose = FALSE)

Arguments

codes codebook matrix for the X space: one row per unit.
data training data, usually.
max.iter default is 20, but convergence usually is much faster. If more iterations are needed, a warning is printed.
verbose print more info to the screen.

Value

Returns a list with components

codes new codebook matrix.
classif classification vector.
niter number of iterations needed for kmeans.

Author(s)

Ron Wehrens


[Package kohonen version 1.0.1 Index]