clusterValidate

Cluster validation

Contents

Syntax

Description

clusterValidate(data, maxClusterNum, opt) returns the best number of clusters based on cluster validation.

• data: Data for clustering, with each column being an entry
• maxClusterNum: Maximum no. of clusters
• opt: options for the function
  • opt.trialNum: no. of trials for each k-means clustering
  • opt.plot: 1 for plotting the result
• bestClusterNum: the returned best number of clusters
• distortion: distortion for each k-means from cluster number from 2 to maxClusterNum.

Example

DS=dcData(6);
maxClusterNum=8;
opt.trialNum=10;
opt.plot=1;
bestClusterNum=clusterValidate(DS.input, maxClusterNum, opt);

Published with MATLAB® R2018a