| 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 | |
; 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))
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 |