kmean.java

一个用于排队系统仿真的开源软件,有非常形象的图象仿真过程!

package jmt.engine.jwat.workloadAnalysis.clustering.kMean;

import java.util.Vector;

import jmt.engine.jwat.MatrixOsservazioni;
import jmt.engine.jwat.workloadAnalysis.clustering.Clustering;
import jmt.engine.jwat.workloadAnalysis.clustering.ClusteringInfos;
import jmt.engine.jwat.workloadAnalysis.clustering.kMean.KMeanClusteringEngine.TempClusterStatistics;
import jmt.gui.jwat.JWATConstants;


public class KMean implements Clustering,JWATConstants{

	private ClusteringInfosKMean[] results;
	private short[][] clustAssign = null;
	//private Vector clusterAssignemnt=null; //Vector di Vector
	private int[] varSel;
	
	public KMean(int numClust,int[] varSel){
		this.varSel=varSel;
		results=new ClusteringInfosKMean[numClust];
		clustAssign=new short[numClust][];
		//clusterAssignemnt= new Vector();
	}
	
	public String getName() {
		return "k-Means";
	}

	public int getNumCluster() {
		return results.length;
	}

	public ClusteringInfos getClusteringInfos(int numCluster) {
		return results[numCluster];
	}
	
	public void calcClusteringInfo(int numCluster,TempClusterStatistics[][] sum,short[] cAss,MatrixOsservazioni m)
	{
		//clusterAssignemnt.add(clusAssign);
		clustAssign[numCluster] = cAss;
		results[numCluster]=new ClusteringInfosKMean(numCluster,m.getNumVariables());
		if(numCluster != 0)
			results[numCluster].Output(varSel,sum,cAss,m,results[numCluster-1].passw);
		else
			results[numCluster].Output(varSel,sum,cAss,m,0);
	}
	
	public void setRatio(int endClust)
	{
		if ( endClust > 2) {
			//Mostra riassunto sui cluster
			/* Calcola un indice che mostri la validit