View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      http://www.apache.org/licenses/LICENSE-2.0
10   *
11   * Unless required by applicable law or agreed to in writing, software
12   * distributed under the License is distributed on an "AS IS" BASIS,
13   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   * See the License for the specific language governing permissions and
15   * limitations under the License.
16   */
17  
18  package org.apache.commons.math4.legacy.ml.clustering.evaluation;
19  
20  import java.util.List;
21  
22  import org.apache.commons.math4.legacy.ml.clustering.Cluster;
23  import org.apache.commons.math4.legacy.ml.clustering.Clusterable;
24  import org.apache.commons.math4.legacy.ml.clustering.ClusterEvaluator;
25  import org.apache.commons.math4.legacy.ml.distance.DistanceMeasure;
26  import org.apache.commons.math4.legacy.stat.descriptive.moment.Variance;
27  
28  /**
29   * Computes the sum of intra-cluster distance variances according to the formula:
30   * <pre>
31   * \( score = \sum\limits_{i=1}^n \sigma_i^2 \)
32   * </pre>
33   * where n is the number of clusters and \( \sigma_i^2 \) is the variance of
34   * intra-cluster distances of cluster \( c_i \).
35   *
36   * @since 3.3
37   */
38  public class SumOfClusterVariances implements ClusterEvaluator {
39      /** The distance measure to use when evaluating the cluster. */
40      private final DistanceMeasure measure;
41  
42      /**
43       * @param measure Distance measure.
44       */
45      public SumOfClusterVariances(final DistanceMeasure measure) {
46          this.measure = measure;
47      }
48  
49      /** {@inheritDoc} */
50      @Override
51      public double score(List<? extends Cluster<? extends Clusterable>> clusters) {
52          double varianceSum = 0.0;
53          for (final Cluster<? extends Clusterable> cluster : clusters) {
54              if (!cluster.getPoints().isEmpty()) {
55  
56                  final Clusterable center = cluster.centroid();
57  
58                  // compute the distance variance of the current cluster
59                  final Variance stat = new Variance();
60                  for (final Clusterable point : cluster.getPoints()) {
61                      stat.increment(distance(point, center));
62                  }
63  
64                  varianceSum += stat.getResult();
65              }
66          }
67          return varianceSum;
68      }
69  
70      /** {@inheritDoc} */
71      @Override
72      public boolean isBetterScore(double a,
73                                   double b) {
74          return a < b;
75      }
76  
77      /**
78       * Calculates the distance between two {@link Clusterable} instances
79       * with the configured {@link DistanceMeasure}.
80       *
81       * @param p1 the first clusterable
82       * @param p2 the second clusterable
83       * @return the distance between the two clusterables
84       */
85      private double distance(final Clusterable p1, final Clusterable p2) {
86          return measure.compute(p1.getPoint(), p2.getPoint());
87      }
88  }