mainfuzzykmean.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.fuzzyKMean;

import java.lang.reflect.InvocationTargetException;

import jmt.engine.jwat.MatrixOsservazioni;
import jmt.engine.jwat.TimeConsumingWorker;
import jmt.engine.jwat.VariableNumber;
import jmt.engine.jwat.workloadAnalysis.clustering.EventClusteringDone;
import jmt.gui.jwat.input.EventFinishAbort;
import jmt.gui.jwat.input.ProgressShow;

public class MainFuzzyKMean extends TimeConsumingWorker {
	
	private MatrixOsservazioni matrix;
	private int maxClust;
	private int maxIter;
	private int fuzzyLevel;
	private int[] varSel;
	private short trasf;
	private FuzzyKMean clustering = null;
	private FuzzyKMeanClusteringEngine cluster = null;
	private String msg=null;
	
	public MainFuzzyKMean(ProgressShow prg,MatrixOsservazioni m,int[] varSel,int numClust,int iteration,int fuzzyL,short trasf) {
		super(prg);
		this.varSel = varSel;
		this.matrix = m;
		this.maxClust = numClust;
		this.maxIter = iteration;
		this.fuzzyLevel = fuzzyL;
		this.trasf = trasf;
		clustering = new FuzzyKMean(maxClust,varSel);
		cluster = new FuzzyKMeanClusteringEngine(clustering,this);
	}
	
	public Object construct() {
		boolean anti = false;
		try {
			initShow((maxIter*maxClust)+3);
			updateInfos(1,"Initializing FuzzyKMeans Clustering",true);
		} catch (InterruptedException e) {
			e.printStackTrace();
		} catch (InvocationTargetException e) {
			e.printStackTrace();
		}
		//Applicazione trasformazione alle variabili coinvolte nel clustering
		if(trasf != VariableNumber.NONE){
			for(int i = 0;i<varSel.length;i++){
				matrix.getVariables()[varSel[i]].doClusteringTrasformation(trasf);
			}
		}
		if (isCanceled()){
			msg="CLUSTERING ABORTED BY USER";
			if(trasf != VariableNumber.NONE){	
				for(int i = 0;i<varSel.length;i++){
					matrix.getVariables()[varSel[i]].undoClueringTrasformation();
				}
			}
			return null;
		}
		try{
			/* Preparazione infos */
			cluster.PrepFClustering(matrix,varSel,maxClust,fuzzyLevel,maxIter);
			if (isCanceled()){
				msg="CLUSTERING ABORTED BY USER";
				if(trasf != VariableNumber.NONE){	
					for(int i = 0;i<varSel.length;i++){
						matrix.getVariables()[varSel[i]].undoClueringTrasformation();
					}
				}
				return null;
			}
			cluster.DoFClustering();
			if (isCanceled()){
				msg="CLUSTERING ABORTED BY USER";
				if(trasf != VariableNumber.NONE){	
					for(int i = 0;i<varSel.length;i++){
						matrix.getVariables()[varSel[i]].undoClueringTrasformation();
					}
				}
				return null;
			}
			if(trasf != VariableNumber.NONE){
				anti = true;
				for(int i = 0;i<varSel.length;i++){
					matrix.getVariables()[varSel[i]].undoClueringTrasformation();
				}
			}
			updateInfos((maxIter*maxClust)+2,"Saving Results",true);
			//Calcolo delle statistiche clustering eseguito
			//for(int i = 0; i < clustering.getNumCluster();i++)
			//clustering.getClusteringInfos(i).DOStat(varSel,clustering.getAsseg()[i],matrix);
			updateInfos((maxIter*maxClust)+3,"END",true);	
			
		}catch(OutOfMemoryError err){
			updateInfos((maxIter*maxClust)+3,"errore",false);
			msg = "Out of Memory. Try with more memory (1Gb JMT Version)";
			if(trasf != VariableNumber.NONE && !anti){	
				for(int i = 0;i<varSel.length;i++){
					matrix.getVariables()[varSel[i]].undoClueringTrasformation();
				}
			}
			return null;
		}
		return clustering;
	}

	public void finished() {
		if(this.get()!=null){
			fireEventStatus(new EventClusteringDone(clustering));
		}
		else{
			fireEventStatus(new EventFinishAbort(msg));
		}
	}

}