k-means算法
发布时间
阅读量：
阅读量
该算法[()是一种经典的无监督学习方法
import java.util.*;
public class KMeans {
private int k;
private double[][] points;
private int[] labels;
private double[][] centroids;
public KMeans(int k, double[][] points) {
this.k = k;
this.points = points;
this.labels = new int[points.length];
this.centroids = new double[k][points[0].length];
// 随机初始化质心
Random rand = new Random();
for (int i = 0; i < k; i++) {
int randomIndex = rand.nextInt(points.length);
centroids[i] = points[randomIndex];
}
}
public void run(int maxIterations) {
for (int iteration = 0; iteration < maxIterations; iteration++) {
// 分配点到最近的质心
assignPointsToClusters();
// 更新质心
updateCentroids();
// 检查是否收敛
if (hasConverged()) {
break;
}
}
}
private void assignPointsToClusters() {
for (int i = 0; i < points.length; i++) {
double minDistance = Double.MAX_VALUE;
int clusterIndex = -1;
for (int j = 0; j < k; j++) {
double distance = calculateDistance(points[i], centroids[j]);
if (distance < minDistance) {
minDistance = distance;
clusterIndex = j;
}
}
labels[i] = clusterIndex;
}
}
private void updateCentroids() {
for (int i = 0; i < k; i++) {
List<double[]> clusterPoints = new ArrayList<>();
for (int j = 0; j < points.length; j++) {
if (labels[j] == i) {
clusterPoints.add(points[j]);
}
}
double[] newCentroid = new double[points[0].length];
for (double[] point : clusterPoints) {
for (int dim = 0; dim < point.length; dim++) {
newCentroid[dim] += point[dim];
}
}
for (int dim = 0; dim < newCentroid.length; dim++) {
newCentroid[dim] /= clusterPoints.size();
}
centroids[i] = newCentroid;
}
}
private boolean hasConverged() {
double[][] newCentroids = new double[k][points[0].length];
for (int i = 0; i < k; i++) {
newCentroids[i] = centroids[i].clone();
}
updateCentroids();
for (int i = 0; i < k; i++) {
if (!Arrays.deepEquals(newCentroids[i], centroids[i])) {
return false;
}
}
return true;
}
private double calculateDistance(double[] point1, double[] point2) {
double sum = 0;
for (int i = 0; i < point1.length; i++) {
sum += Math.pow(point1[i] - point2[i], 2);
}
return Math.sqrt(sum);
}
public void printClusters() {
for (int i = 0; i < k; i++) {
System.out.println("Cluster " + i + ":");
for (int j = 0; j < points.length; j++) {
if (labels[j] == i) {
System.out.println(Arrays.toString(points[j]));
}
}
System.out.println("Centroid: " + Arrays.toString(centroids[i]));
System.out.println();
}
}全部评论 (0)
还没有任何评论哟~