Java实现聚类算法k-means

  1 import java.sql.Array;
  2 import java.util.ArrayList;
  3 import java.util.Random;
  4 import java.util.Scanner;
  5 
  6 class point {
  7     public float x = 0;
  8     public float y = 0;
  9     public int flage = -1;
 10 
 11     public float getX() {
 12         return x;
 13     }
 14 
 15     public void setX(float x) {
 16         this.x = x;
 17     }
 18 
 19     public float getY() {
 20         return y;
 21     }
 22 
 23     public void setY(float y) {
 24         this.y = y;
 25     }
 26 }
 27 
 28 public class Kcluster {
 29 
 30     point[] ypo;// 点集
 31     point[] pacore = null;// old聚类中心
 32     point[] pacoren = null;// new聚类中心
 33 
 34     // 初试聚类中心，点集
 35     public void productpoint() {
 36         Scanner cina = new Scanner(System.in);
 37         System.out.print("请输入聚类中点的个数（随机产生）：");
 38         int num = cina.nextInt();
 39 
 40         ypo = new point[num];
 41         // 随机产生点
 42         for (int i = 0; i < num; i++) {
 43 
 44             float x = (int) (new Random().nextInt(10));
 45             float y = (int) (new Random().nextInt(10));
 46 
 47             ypo[i] = new point();// 对象创建
 48             ypo[i].setX(x);
 49             ypo[i].setY(y);
 50 
 51         }
 52 
 53         // 初始化聚类中心位置
 54         System.out.print("请输入初始化聚类中心个数（随机产生）：");
 55         int core = cina.nextInt();
 56         this.pacore = new point[core];// 存放聚类中心
 57         this.pacoren = new point[core];
 58 
 59         Random rand = new Random();
 60         int temp[] = new int[core];
 61         temp[0] = rand.nextInt(num);
 62         pacore[0] = new point();
 63         pacore[0] = ypo[temp[0]];
 64         // 避免产生重复的中心
 65         for (int i = 1; i < core; i++) {
 66             int flage = 0;
 67             int thistemp = rand.nextInt(num);
 68             for (int j = 0; j < i; j++) {
 69                 if (temp[j] == thistemp) {
 70                     flage = 1;// 有重复
 71                     break;
 72 
 73                 }
 74             }
 75             if (flage == 1) {
 76                 i--;
 77             } else {
 78                 pacore[i] = new point();
 79                 pacore[i] = ypo[thistemp];
 80                 pacore[i].flage = 0;// 0表示聚类中心
 81             }
 82 
 83         }
 84         System.out.println("初始聚类中心：");
 85         for (int i = 0; i < pacore.length; i++) {
 86             System.out.println(pacore[i].x + " " + pacore[i].y);
 87         }
 88 
 89     }
 90 
 91     // ///找出每个点属于哪个聚类中心
 92     public void searchbelong()// 找出每个点属于哪个聚类中心
 93     {
 94 
 95         for (int i = 0; i < ypo.length; i++) {
 96             double dist = 999;
 97             int lable = -1;
 98             for (int j = 0; j < pacore.length; j++) {
 99 
100                 double distance = distpoint(ypo[i], pacore[j]);
101                 if (distance < dist) {
102                     dist = distance;
103                     lable = j;
104                     // po[i].flage = j + 1;// 1,2,3......
105 
106                 }
107             }
108             ypo[i].flage = lable + 1;
109 
110         }
111 
112     }
113 
114     // 更新聚类中心
115     public void calaverage() {
116 
117         for (int i = 0; i < pacore.length; i++) {
118             System.out.println("以<" + pacore[i].x + "," + pacore[i].y
119                     + ">为中心的点：");
120             int numc = 0;
121             point newcore = new point();
122             for (int j = 0; j < ypo.length; j++) {
123 
124                 if (ypo[j].flage == (i + 1)) {
125                     numc += 1;
126                     newcore.x += ypo[j].x;
127                     newcore.y += ypo[j].y;
128                     System.out.println(ypo[j].x + "," + ypo[j].y);
129                 }
130             }
131             // 新的聚类中心
132             pacoren[i] = new point();
133             pacoren[i].x = newcore.x / numc;
134             pacoren[i].y = newcore.y / numc;
135             pacoren[i].flage = 0;
136             System.out.println("新的聚类中心：" + pacoren[i].x + "," + pacoren[i].y);
137 
138         }
139     }
140 
141     public double distpoint(point px, point py) {
142 
143         return Math.sqrt(Math.pow((px.x - py.x), 2)
144                 + Math.pow((px.y - py.y), 2));
145 
146     }
147 
148     public void change_oldtonew(point[] old, point[] news) {
149         for (int i = 0; i < old.length; i++) {
150             old[i].x = news[i].x;
151             old[i].y = news[i].y;
152             old[i].flage = 0;// 表示为聚类中心的标志。
153         }
154     }
155 
156     public void movecore() {
157         // this.productpoint();//初始化，样本集，聚类中心，
158         this.searchbelong();
159         this.calaverage();//
160         double movedistance = 0;
161         int biao = -1;//标志，聚类中心点的移动是否符合最小距离
162         for (int i = 0; i < pacore.length; i++) {
163             movedistance = distpoint(pacore[i], pacoren[i]);
164             System.out.println("distcore:" + movedistance);//聚类中心的移动距离
165             if (movedistance < 0.01) {
166                 biao = 0;
167 
168             } else {
169                 
170                 biao=1;
171                 break;
172 
173             }
174         }
175         if (biao == 0) {
176             System.out.print("迭代完毕！！！！！");
177         } else {
178             change_oldtonew(pacore, pacoren);
179             movecore();
180         }
181 
182     }
183 
184     public static void main(String[] args) {
185         // TODO Auto-generated method stub
186 
187         Kcluster kmean = new Kcluster();
188         kmean.productpoint();
189         kmean.movecore();
190     }
191 
192 }