Kmeans思想

Kmeans(K均值算法):

 

代码:import java.util.ArrayList;import java.util.Random;public class Kmeans {private int k;// 参数kprivate int m;// 迭代次数mprivate int dataSetLength; // 数据集长度，即数据集个数private ArrayList<Float[]> dataSet; // 数据集链表private ArrayList<Float[]> center;// 中心链表private ArrayList<ArrayList<Float[]>> cluster;private ArrayList<Float> jc;// 误差平方合private Random random;public Kmeans(int k) {this.k = k;}public void init() {m = 0;random = new Random();dataSet = initDataSet();// 初始化数据集}/** * 初始化数据集方法 * @return */public ArrayList<Float[]> initDataSet() {ArrayList<Float[]> dataSet = new ArrayList<Float[]>();//15Float[][] dataSetArray = new Float[][] {{8f, 2f}, {3f, 4f}, {2f, 5f}, {4f, 2f}, {7f, 3f},{6f, 2f}, {4f, 7f}, {6f, 3f}, {5f, 3f}, {6f, 3f},{6f, 9f}, {1f, 6f}, {3f, 9f}, {4f, 1f}, {8f, 6f}};dataSetLength=dataSetArray.length;for(int i = 0; i<dataSetArray.length;i++){dataSet.add(dataSetArray[i]);}return dataSet;}/** * 初始化中心方法 */public ArrayList<Float[]> initCenter(){ArrayList<Float[]> center = new ArrayList<Float[]>();//给center分配内存int[] randoms = new int[k];//定义一个一维数组randoms,数组length=3boolean flag ;int temp = random.nextInt(dataSetLength);//随机生成一个0到dataSetLength的int型数值,此处为0~15randoms[0] = temp;//把随机数传给randoms[0]for(int i = 1;i <k;i++)//执行2次{flag = true;while (flag){temp=random.nextInt(dataSetLength);//生成一个0~15的随机数交给tempint j;for(j=0;j<i;j++){if(temp == randoms[j])//如果此随机数等于randoms[0]{break;//跳出循环}}if(j==i){flag=false;}}randoms[i] = temp;}/*//测试随机数组生成情况for(int i = 0; i < k; i ++){System.out.println("test1: randoms[" + i + "] = " + randoms[i]);}System.out.println();*/for (int i = 0; i < k; i ++){center.add(dataSet.get(randoms[i])); //生成初始化中心链表}return center;}public ArrayList<Float> initJc(){ArrayList<Float> jc = new ArrayList<Float>();return jc;}/* * 初始化簇集方法 */public ArrayList<ArrayList<Float[]>> initCluster(){ArrayList<ArrayList<Float[]>> cluster = new ArrayList<ArrayList<Float[]>>();for(int i = 0; i < k; i ++){cluster.add(new ArrayList<Float[]>());}return cluster;}/* * 求距离方法 */public float distance(Float[] element, Float[] center){float distance = 0.0f;float x = element[0] - center[0];float y = element[1] - center[1];float z = x*x + y*y;distance = (float) Math.sqrt(z);return distance;}/* * 求最小距离位置方法 */public int minDistance(float distance[]){float minDistance = distance[0];int minLocation = 0;for(int i = 1; i < distance.length; i ++){if(distance[i] < minDistance){minDistance = distance[i];minLocation = i;}else if(distance[i] == minDistance){ //如果和当前最短距离相等，则随机选取一个if(random.nextInt(10) < 5){minLocation = i;}}}return minLocation;}/** * 生成簇集元素方法 */public void clusterSet(){float[] distance = new float[k];for(int i = 0; i < dataSetLength; i ++){for(int j = 0; j < k; j ++){distance[j] =distance(dataSet.get(i), center.get(j));//System.out.println("test2: " + "dataSet[" + i + "], centers[" + j + "], distance = " + distance[j]); //测试元素与中心距离}int minLocation = minDistance(distance);cluster.get(minLocation).add(dataSet.get(i)); //核心：将当前元素放到最小距离中心相关的簇中}}/* * 求误差平方的方法 */public float errorSquare(Float[] element, Float[] center){float x = element[0] - center[0];float y = element[1] - center[1];float errorSquare = x*x + y*y;return errorSquare;}/** * 计算误差平方和准则函数方法 */public void countRule(){float JcF = 0;for(int i = 0; i < cluster.size(); i ++){//即i<3for(int j = 0; j < cluster.get(i).size(); j ++){//执行的第i个簇的长度的次数JcF += errorSquare(cluster.get(i).get(j), center.get(i));//jcf=jcf+(聚类的点x-质心x)的平方+(聚类的点y-质心y)的平方}}jc.add(JcF);}/** * 计算新的簇中心方法 */public void findNewCenter(){for(int i = 0; i < k; i ++){//k=3依次运算3个质心int n = cluster.get(i).size();//得到第i个簇的数量 (n=5),(一共3个蔟)if(n != 0){//如果该蔟的点的个书不为空Float[] newCenter = {0.0f, 0.0f};//声明一个新的质心点（默认坐标为0,0）for(int j = 0; j < n; j ++){//依次对n个点做循环newCenter[0] += cluster.get(i).get(j)[0];//新的质心的x坐标为第i个蔟的所有点的x坐标的和newCenter[1] += cluster.get(i).get(j)[1];//新的质心的y坐标为第i个蔟的所有点的y坐标的和}newCenter[0] = newCenter[0] / n;//求质心newCenter[1] = newCenter[1] / n;//求质心center.set(i, newCenter);//重新设第i个蔟的质心为新的newCenter}}}/* * 打印数据数组 */public void printDataArray(ArrayList<Float[]> dataArray, String dataArrayName){for(int i = 0; i < dataArray.size(); i ++){System.out.println("print: " + dataArrayName + "[" + i + "] = {" + dataArray.get(i)[0] + ", " + dataArray.get(i)[1] + "}");}System.out.println();}/* * Kmeans算法核心过程方法 */public void kmeans(){init(); //初始化center=initCenter();cluster=initCluster();jc=initJc();printDataArray(dataSet, "initDataSet"); //输出初始化数据集printDataArray(center, "initCenter"); //输出初始化中心while(true){clusterSet(); //生成簇集元素for (int i = 0; i < cluster.size(); i++) {printDataArray(cluster.get(i), "cluster[" + i + "]"); //输出簇集生成结果}countRule(); //计算误差平方和System.out.println("count:" + "Jc[" + m + "] = " + jc.get(m)); //输出误差平方和System.out.println();//判断退出迭代条件if(m != 0){//如果m不为0而且jc的变化为0则跳出循环if (jc.get(m) - jc.get(m - 1) == 0) {break;}}findNewCenter(); //计算新的中心printDataArray(center, "newCenter"); //输出新的中心m++;//m+1cluster.clear(); //簇集清空cluster = initCluster(); //簇集初始化}System.out.println("note: the times of repeat: m = " + m); //输出迭代次数}/* * 主函数 */public static void main(String[] args){long startTime = System.currentTimeMillis(); //获取开始时间System.out.println("note: program begins.");Kmeans myKmeans = new Kmeans(3);myKmeans.kmeans(); //调用Kmeans核心方法long endTime = System.currentTimeMillis(); //获取结束时间System.out.println("note: running time = " +(endTime - startTime) + "ms.");System.out.println("note: program ends.");}}