聚類算法之DBScan(Java實現)[轉]

package orisun;
 
import java.io.File;
import java.util.ArrayList;
import java.util.Vector;
import java.util.Iterator;
 
public class DBScan {
 
    double Eps=3;   //區域半徑
    int MinPts=4;   //密度
     
    //因爲本身到本身的距離是0,因此本身也是本身的neighbor
    public Vector<DataObject> getNeighbors(DataObject p,ArrayList<DataObject> objects){
        Vector<DataObject> neighbors=new Vector<DataObject>();
        Iterator<DataObject> iter=objects.iterator();
        while(iter.hasNext()){
            DataObject q=iter.next();
            double[] arr1=p.getVector();
            double[] arr2=q.getVector();
            int len=arr1.length;
             
            if(Global.calEditDist(arr1,arr2,len)<=Eps){      //使用編輯距離
//          if(Global.calEuraDist(arr1, arr2, len)<=Eps){    //使用歐氏距離    
//          if(Global.calCityBlockDist(arr1, arr2, len)<=Eps){   //使用街區距離
//          if(Global.calSinDist(arr1, arr2, len)<=Eps){ //使用向量夾角的正弦
                neighbors.add(q);
            }
        }
        return neighbors;
    }
     
    public int dbscan(ArrayList<DataObject> objects){
        int clusterID=0;
        boolean AllVisited=false;
        while(!AllVisited){
            Iterator<DataObject> iter=objects.iterator();
            while(iter.hasNext()){
                DataObject p=iter.next();
                if(p.isVisited())
                    continue;
                AllVisited=false;
                p.setVisited(true);     //設爲visited後就已經肯定了它是核心點仍是邊界點
                Vector<DataObject> neighbors=getNeighbors(p,objects);
                if(neighbors.size()<MinPts){
                    if(p.getCid()<=0)
                        p.setCid(-1);       //cid初始爲0,表示未分類；分類後設置爲一個正數；設置爲-1表示噪聲。
                }else{
                    if(p.getCid()<=0){
                        clusterID++;
                        expandCluster(p,neighbors,clusterID,objects);
                    }else{
                        int iid=p.getCid();
                        expandCluster(p,neighbors,iid,objects);
                    }
                }
                AllVisited=true;
            }
        }
        return clusterID;
    }
 
    private void expandCluster(DataObject p, Vector<DataObject> neighbors,
            int clusterID,ArrayList<DataObject> objects) {
        p.setCid(clusterID);
        Iterator<DataObject> iter=neighbors.iterator();
        while(iter.hasNext()){
            DataObject q=iter.next();
            if(!q.isVisited()){
                q.setVisited(true);
                Vector<DataObject> qneighbors=getNeighbors(q,objects);
                if(qneighbors.size()>=MinPts){
                    Iterator<DataObject> it=qneighbors.iterator();
                    while(it.hasNext()){
                        DataObject no=it.next();
                        if(no.getCid()<=0)
                            no.setCid(clusterID);
                    }
                }
            }
            if(q.getCid()<=0){       //q不是任何簇的成員
                q.setCid(clusterID);
            }
        }
    }
 
    public static void main(String[] args){
        DataSource datasource=new DataSource();
        //Eps=3,MinPts=4
        datasource.readMatrix(new File("/home/orisun/test/dot.mat"));
        datasource.readRLabel(new File("/home/orisun/test/dot.rlabel"));
        //Eps=2.5,MinPts=4
//      datasource.readMatrix(new File("/home/orisun/text.normalized.mat"));
//      datasource.readRLabel(new File("/home/orisun/text.rlabel"));
        DBScan ds=new DBScan();
        int clunum=ds.dbscan(datasource.objects);
        datasource.printResult(datasource.objects,clunum);
    }
}