GMD (version 0.3.3)

css: Clustering Sum-of-Squares for clustering evaluation

Description

Evaluation on the varaince of a clustering model using squared Euclidean distances, based on distance matrix and cluster membership.

Usage

css(dist.obj,clusters)
## Computing Sum-of-Squares given Hierarchical Clustering
"css"(dist.obj, hclust.obj=NULL,hclust.FUN=hclust,
hclust.FUN.MoreArgs=list(method="ward"),k=NULL)

Arguments

dist.obj

a `dist' object as produced by dist or gdist.

clusters

a vector with cluster memberships.

k

numeric, the upper bound of the number of clusters to compute. DEFAULT: 20 or the number of observations (if less than 20).

hclust.obj

a `hclust' object, generated by hclust

hclust.FUN

a function, to generate a hierarchical clustering. Ignored with hclust.obj specified. DEFAULT: hclust

hclust.FUN.MoreArgs

a list, containing arguments that are passed to hclust.FUN.

Value

css returns a `css' object, which is a list containing the following components

k

number of clusters

wss

k within-cluster sum-of-squares

totwss

total within-cluster sum-of-square

totbss

total between-cluster sum-of-square

tss

total sum of squares of the data

, and with an attribute `meta' that contains the input components

dist.obj

(the input) distance matrix

clusters

(the input) cluster membership

css.hclust returns a `css.multi' object, which is a data.frame containing the following columns

k

number of clusters

ev

explained variance given k

totbss

total between-cluster sum-of-square

tss

total sum of squares of the data

, and with an attribute `meta' that contains

cmethod

the clustering method

dist.obj

(the input) distance matrix

k

(the input) number of clusters

clusters

the `hclust' object that is either by input or computed by default

Details

Clustering Sum-of-Squares for clustering evaluation.

See Also

elbow for "elbow" plot using `css.multi' object