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: kmeans adaptive agglom

Quantlet: divisive
Description: performs an adaptive divisive K-means cluster analysis with appropriate (adaptive) multivariate graphic using principal components

Usage: cd = divisive (x, k, w, m, sv)
Input:
x n x p matrix of n row points to be clustered
k scalar number of clusters
w p x 1 matrix of weights of column points
m n x 1 matrix of weights (masses) of row points
sv scalar seed value for random numbers
Output:
cd.p n x 1 matrix partition of n points of x into k clusters
cd.n k x 1 matrix of number of observations of clusters
cd. a p x 1 matrix of final (pooled) adaptive weights of the variables

Example:

; load the library xclust

library ("xclust")

; initialize random generator

randomize(0)

; generate basis data 

x  = normal(30, 5)

; generate 4 clusters

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

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

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

x4 = x - #(0,2,1,3,0)'

x  = x1|x2|x3|x4

; compute column variances

w  = 1./var(x)

; compute row weights

m  = matrix(rows(x))

; apply divisive

cd = divisive (x, 4, w, m, 1111)

; compare estimated and true partition

conting (cd.p, 1+ceil((1:120)/30))

Result:

Content of object h

[1,]        0       30        0        0 

[2,]        0        0       30        0 

[3,]       30        0        0        0 

[4,]        0        0        0       30 


Library: xclust
See also: kmeans adaptive agglom

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