Java实现数据挖掘K-Means算法

一、环境:

    Windows10，Eclipse，Java JDK1.8(1.7以上都可以)

二、数据集（部分）

iris.txt 1.4,0.2,Iris-setosa 1.4,0.2,Iris-setosa 1.3,0.2,Iris-setosa 1.5,0.2,Iris-setosa 1.4,0.2,Iris-setosa 1.7,0.4,Iris-setosa 1.4,0.3,Iris-setosa 1.5,0.2,Iris-setosa 1.4,0.2,Iris-setosa 1.5,0.1,Iris-setosa 1.5,0.2,Iris-setosa 1.6,0.2,Iris-setosa 1.4,0.1,Iris-setosa ....... ....... .......

三、代码 

package kmeans;

import java.io.*;
import java.nio.*;
import java.nio.file.*;
import java.util.*;

public class KMeans {
	//一条记录
	private class Record{
		int id;
		double x,y;
		int meanId;
		Record(int id,double x,double y){
			this.id = id;this.x=x;this.y=y;
		}
	}
	//K值
	private final int K=3;
	//结束算法的条件，也就是新的均值点与旧的均值点所允许的最小距离
	private final double DIS = 0.01;
	//保存所有记录的列表
	List<Record> recordList = new ArrayList<>();
	//记录的总数
	int recNum;
	//均值点列表
	List<Record> meansList = new ArrayList<>(K);
	//保存聚类后的记录的列表，其本质是对recordList中记录的引用
	List<List<Record>> kRecordList = new ArrayList<>(K);
	
	/**
	 * 构造函数
	 */
	public KMeans(String file){
		ReadAllRecords(file);
		recNum = recordList.size();
		initKMeans();
		K_Means();
	}
	
	/**
	 * 读取所有记录行的函数
	 * file是文件位置
	 */
	private void ReadAllRecords(String file){
		try {
			Path path = Paths.get(file);
			BufferedReader br= Files.newBufferedReader(path);
			String line = "";
			String []s = null;
			int i=0;
			while((line=br.readLine())!=null){
				s = line.split(",");
				Record record = new Record(i,Double.parseDouble(s[0]),
						Double.parseDouble(s[1]));
				recordList.add(record);
				i++;
			}
		} catch (IOException e) {
			// TODO 自动生成的 catch 块
			e.printStackTrace();
		}
	}
	
	/**
	 * 计算欧式距离
	 * @param mean  均值点
	 */
	private double Distance(Record mean,Record r){
		return Math.sqrt((mean.x-r.x)*(mean.x-r.x)+(mean.y-r.y)*(mean.y-r.y));
	}
	
	/**
	 * 构造初始的K个均值点,并分配所有记录所属的类别
	 */
	private void initKMeans() {
		int k = recNum/K;
		//选K个均值点，其距离为k
		for(int i=0;i<K;i++) {
			Record r = new Record(i,recordList.get(i*k).x,recordList.get(i*k).y);
			meansList.add(r);
		}
		//初始化kRecordsList
		for(int i=0;i<K;i++) {
			kRecordList.add(new ArrayList<Record>());
		}
		//为所有点聚类
		int minMeanId;//距离点最近的距离的均值点点ID
		double minDist;//最短距离
		double dist;
		for(int i=0;i<recNum;i++) {
			minMeanId = meansList.get(0).id;
			minDist = Distance(meansList.get(0),recordList.get(i));
			for(int j=1;j<K;j++) {
				dist = Distance(meansList.get(j),recordList.get(i));
				if(dist<minDist) {
					minDist = dist;
					minMeanId = meansList.get(j).id;
				}
			}
			recordList.get(i).meanId=minMeanId;
			kRecordList.get(minMeanId).add(recordList.get(i));
		}
	}
	
	/**
	 * K-Means算法函数
	 */
	public void K_Means(){
		//循环执行300伺候强制结束算法
		int n=0;
		while(true) {
			if(300<=n++) break;
			//首先生成新的K个均值点
			Record mean = null;
			double distsum;
			int z =0;//均值点与上次均值点相差不大的点的数量
			for(int i=0;i<K;i++) {
				distsum = 0;
				mean = new Record(i,0,0);
				for(int j=0;j<kRecordList.get(i).size();j++) {
					mean.x+=kRecordList.get(i).get(j).x;
					mean.y+=kRecordList.get(i).get(j).y;
				}
				mean.x/=kRecordList.get(i).size();
				mean.y/=kRecordList.get(i).size();
				if(Distance(meansList.get(i),mean)<=DIS) {
					z++;
				}
				meansList.set(i,mean);
			}
			if(z==3) break;
			//清空kRecordsList的所有列
			for(int i=0;i<K;i++) {
				kRecordList.get(i).clear();;
			}
			//然后重新进行聚类
			int minMeanId;//距离点最近的距离的均值点点ID
			double minDist;//最短距离
			double dist;
			for(int i=0;i<recNum;i++) {
				minMeanId = meansList.get(0).id;
				minDist = Distance(meansList.get(0),recordList.get(i));
				for(int j=1;j<K;j++) {
					dist = Distance(meansList.get(j),recordList.get(i));
					if(dist<minDist) {
						minDist = dist;
						minMeanId = meansList.get(j).id;
					}
				}
				recordList.get(i).meanId=minMeanId;
				kRecordList.get(minMeanId).add(recordList.get(i));
			}
		}
	}
	public void ShowRecords() {
		Iterator<List<Record>> iList = kRecordList.iterator();
		int i=0;
		while(iList.hasNext()) {
			System.out.print("均值点:("+meansList.get(i).x+","+meansList.get(i).y+") 成员:");
			Iterator<Record> iRecord = iList.next().iterator();
			i++;
			while(iRecord.hasNext()) {
				Record record = iRecord.next();
				System.out.print("("+record.x+","+record.y+") ");
			}
			System.out.println();
		}
	}
	
	public static void main(String[] args) {
		KMeans kmeans = new KMeans("./files/iris.txt");
		kmeans.ShowRecords();
	}
}