Example usage for org.apache.commons.math3.stat.clustering KMeansPlusPlusClusterer cluster

Introduction

In this page you can find the example usage for org.apache.commons.math3.stat.clustering KMeansPlusPlusClusterer cluster.

Prototype

public List<Cluster<T>> cluster(final Collection<T> points, final int k, final int maxIterations)
        throws MathIllegalArgumentException, ConvergenceException 

Source Link

Document

Runs the K-means++ clustering algorithm.

Usage

From source file:org.moeaframework.core.operator.ParentCentricVariationTest.java

@Override
protected void check(Solution[] parents, Solution[] offspring) {
    List<ClusterablePoint> points = new ArrayList<ClusterablePoint>();

    for (Solution solution : offspring) {
        points.add(new ClusterablePoint(EncodingUtils.getReal(solution)));
    }//from  w w  w. j a  v a 2 s.c  om

    KMeansPlusPlusClusterer<ClusterablePoint> clusterer = new KMeansPlusPlusClusterer<ClusterablePoint>(
            new Random());

    List<Cluster<ClusterablePoint>> clusters = clusterer.cluster(points, parents.length, 100);

    for (Solution solution : parents) {
        boolean match = false;

        for (int i = 0; i < clusters.size(); i++) {
            boolean allEqual = true;

            double[] centroid = clusters.get(i).getCenter().getPoint();
            double[] parent = EncodingUtils.getReal(solution);

            for (int j = 0; j < parent.length; j++) {
                if (Math.abs(parent[j] - centroid[j]) > TestThresholds.VARIATION_EPS) {
                    allEqual = false;
                }
            }

            if (allEqual) {
                match = true;
            }
        }

        Assert.assertTrue(match);
    }
}

---

From source file:org.moeaframework.core.operator.real.AdaptiveMetropolisTest.java

/**
 * Tests if each cluster formed around the parents exhibits the same
 * covariance as the parent solutions./*from www.  jav a2  s . co m*/
 */
@Test
public void testCovariance() {
    //the smaller jump rate is used to ensure separation between clusters
    double jumpRateCoefficient = 0.5;

    AdaptiveMetropolis am = new AdaptiveMetropolis(3, TestThresholds.SAMPLES, jumpRateCoefficient);

    Solution[] parents = new Solution[] { newSolution(0.0, 0.0), newSolution(0.0, 5.0), newSolution(2.0, 0.0) };

    Solution[] offspring = am.evolve(parents);

    List<ClusterablePoint> points = new ArrayList<ClusterablePoint>();

    for (Solution solution : offspring) {
        points.add(new ClusterablePoint(EncodingUtils.getReal(solution)));
    }

    KMeansPlusPlusClusterer<ClusterablePoint> clusterer = new KMeansPlusPlusClusterer<ClusterablePoint>(
            new Random());

    List<Cluster<ClusterablePoint>> clusters = clusterer.cluster(points, parents.length, 100);

    for (Cluster<ClusterablePoint> cluster : clusters) {
        TestUtils.assertEquals(getCovariance(cluster), getCovariance(parents, jumpRateCoefficient),
                new RelativeError(0.1));
    }
}

---