k-means clustering

 

kmeans	k-means clustering
-d <data>	name of the input data frame
-dout <dataout>	name of the output data frame
-cout <cldata>	data frame for clusters
-k <k>	number of k-means clusters

This command performs the k-means clustering algorithm in order to divide data into <k> clusters. The centroids of the clusters are stored into the frame <dataout> and the classification information of data records into the classified data <cldata>.

Example (ex5.15): In the example, the Boston data is divided into five clusters, which are visualized through their centroids.

NDA> load boston.dat
NDA> select flds -f boston.indus boston.age boston.zn boston.dis
NDA> kmeans -d flds -dout clusdat -cout cluscld -k 5
NDA> mkgrp win1 -d clusdat
...