Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: xclust
See also: adaptive

Quantlet: adap
Description: performs an adaptive K-means cluster analysis

Usage: ca = adap(x, k, w, m, t)
Input:
x n x p matrix of n row points to be clustered
k scalar the number of clusters
w p x 1 matrix of weights of column points
m n x 1 matrix of weights (masses) of row points
t n x 1 matrix of the true partition (only if known, else a matrix containing 1)
Output:
ca.b n x 1 matrix partition of n points into k clusters
ca.c k x p matrix of means (centroids) of clusters
ca.v k x p matrix of within cluster variances divided by the corresponding weights (masses) of clusters
ca.s k x 1 matrix of weights (masses) of clusters
ca.a p x 1 matrix of adaptive weights of variables

Example:

; load the library xclust

library ("xclust")

; initialize random generator

randomize(0)

; generate some normal data

x  = normal(200, 5)

x1 = x - #(2,1,3,0,0)' 

x2 = x + #(1,1,3,1,0.5)'                                   

x3 = x + #(0,0,1,5,1)'    

; make one data set

x  = x1|x2|x3

; compute column variances

w  = 1./var(x)

; generate row weights (here : 1)

m  = matrix(rows(x))

; generate true partition

t  = matrix(200)|matrix(200).+1|matrix(200).+2 

; apply adaptive clustering

ca = adap (x, 3, w, m, t)

ca

Result:

gives a partition ca.b of n row points into 3 clusters which 

minimizes the sum of within cluster variances according

to the column weights (1/pooled within cluster variances)


Library: xclust
See also: adaptive

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Author: Hans-Joachim Mucha, 950121 Sigbert Klinke, 970902
(C) MD*TECH Method and Data Technologies, 21.9.2000