1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
| // Kmeans.java
package com.orhandemirel.clustering;
import java.util.Random;
import java.util.ArrayList;
public class Kmeans{
private double[][] data; // data to cluster
private int numClusters; // number of clusters
private double[][] clusterCenters; // cluster centers
private int dataSize; // size of the data
private int dataDim; // dimention of the data
private ArrayList[] clusters; // calculated clusters
private double[] clusterVars; // cluster variances
private double epsilon;
public Kmeans(double[][] data, int numClusters, double[][] clusterCenters)
{
dataSize = data.length;
dataDim = data[0].length;
this.data = data;
this.numClusters = numClusters;
this.clusterCenters = clusterCenters;
clusters = new ArrayList[numClusters];
for(int i=0;i<numClusters;i++)
{
clusters[i] = new ArrayList();
}
clusterVars = new double[numClusters];
epsilon = 0.01;
}
public Kmeans(double[][] data, int numClusters)
{
this(data, numClusters, true);
}
public Kmeans(double[][] data, int numClusters, boolean randomizeCenters)
{
dataSize = data.length;
dataDim = data[0].length;
this.data = data;
this.numClusters = numClusters;
this.clusterCenters = new double[numClusters][dataDim];
clusters = new ArrayList[numClusters];
for(int i=0;i<numClusters;i++)
{
clusters[i] = new ArrayList();
}
clusterVars = new double[numClusters];
epsilon = 0.01;
if(randomizeCenters)
{
randomizeCenters(numClusters, data);
}
}
private void randomizeCenters(int numClusters, double[][] data) {
Random r = new Random();
int[] check = new int[numClusters];
for (int i = 0; i < numClusters; i++) {
int rand = r.nextInt(dataSize);
if (check[i] == 0) {
this.clusterCenters[i] = data[rand].clone();
check[i] = 1;
} else {
i--;
}
}
}
private void calculateClusterCenters()
{
for(int i=0;i<numClusters;i++)
{
int clustSize = clusters[i].size();
for(int k= 0; k < dataDim; k++)
{
double sum = 0d;
for(int j =0; j < clustSize; j ++)
{
double[] elem = (double[]) clusters[i].get(j);
sum += elem[k];
}
clusterCenters[i][k] = sum / clustSize;
}
}
}
private void calculateClusterVars()
{
for(int i=0;i<numClusters;i++)
{
int clustSize = clusters[i].size();
Double sum = 0d;
for(int j =0; j < clustSize; j ++)
{
double[] elem = (double[])clusters[i].get(j);
for(int k= 0; k < dataDim; k++)
{
sum += Math.pow( (Double)elem[k] - getClusterCenters()[i][k], 2);
}
}
clusterVars[i] = sum / clustSize;
}
}
public double getTotalVar()
{
double total = 0d;
for(int i=0;i< numClusters;i++)
{
total += clusterVars[i];
}
return total;
}
public double[] getClusterVars()
{
return clusterVars;
}
public ArrayList[] getClusters()
{
return clusters;
}
private void assignData()
{
for(int k=0;k<numClusters;k++)
{
clusters[k].clear();
}
for(int i=0; i<dataSize; i++)
{
int clust = 0;
double dist = Double.MAX_VALUE;
double newdist = 0;
for(int j=0; j<numClusters; j++)
{
newdist = distToCenter( data[i], j );
if( newdist <= dist )
{
clust = j;
dist = newdist;
}
}
clusters[clust].add(data[i]);
}
}
private double distToCenter( double[] datum, int j )
{
double sum = 0d;
for(int i=0;i < dataDim; i++)
{
sum += Math.pow(( datum[i] - getClusterCenters()[j][i] ), 2);
}
return Math.sqrt(sum);
}
public void calculateClusters()
{
double var1 = Double.MAX_VALUE;
double var2;
double delta;
do
{
calculateClusterCenters();
assignData();
calculateClusterVars();
var2 = getTotalVar();
if (Double.isNaN(var2)) // if this happens, there must be some empty clusters
{
delta = Double.MAX_VALUE;
randomizeCenters(numClusters, data);
assignData();
calculateClusterCenters();
calculateClusterVars();
}
else
{
delta = Math.abs(var1 - var2);
var1 = var2;
}
}while(delta > epsilon);
}
public void setEpsilon(double epsilon)
{
if(epsilon > 0)
{
this.epsilon = epsilon;
}
}
/**
* @return the clusterCenters
*/
public double[][] getClusterCenters() {
return clusterCenters;
}
} |
Implémenter l'algorithme K-means en Java
Répondre avec citation
Partager