Example usage for org.apache.commons.math.stat.clustering Cluster Cluster

Introduction

In this page you can find the example usage for org.apache.commons.math.stat.clustering Cluster Cluster.

Prototype

public Cluster(final T center) 

Source Link

Document

Build a cluster centered at a specified point.

Usage

From source file:org.basketball.MyKMeansPlusPlusClusterer.java

/**
 * Runs the K-means++ clustering algorithm.
 *
 * @param points the points to cluster//from  w w  w. java  2 s  .c om
 * @param k the number of clusters to split the data into
 * @param maxIterations the maximum number of iterations to run the algorithm
 *     for.  If negative, no maximum will be used
 * @return a list of clusters containing the points
 */
public List<Cluster<T>> cluster(final Collection<T> points, final int k, final int maxIterations) {
    // create the initial clusters
    List<Cluster<T>> clusters = chooseInitialCenters(points, k, random);
    assignPointsToClusters(clusters, points);

    // iterate through updating the centers until we're done
    final int max = (maxIterations < 0) ? Integer.MAX_VALUE : maxIterations;
    for (int count = 0; count < max; count++) {
        boolean clusteringChanged = false;
        List<Cluster<T>> newClusters = new ArrayList<Cluster<T>>();
        for (final Cluster<T> cluster : clusters) {
            final T newCenter;
            if (cluster.getPoints().isEmpty()) {
                switch (emptyStrategy) {
                case LARGEST_VARIANCE:
                    newCenter = getPointFromLargestVarianceCluster(clusters);
                    break;
                case LARGEST_POINTS_NUMBER:
                    newCenter = getPointFromLargestNumberCluster(clusters);
                    break;
                case FARTHEST_POINT:
                    newCenter = getFarthestPoint(clusters);
                    break;
                case IGNORE:
                    continue;
                default:
                    throw new ConvergenceException(LocalizedFormats.EMPTY_CLUSTER_IN_K_MEANS);
                }
                clusteringChanged = true;
            } else {
                newCenter = cluster.getCenter().centroidOf(cluster.getPoints());
                if (!newCenter.equals(cluster.getCenter())) {
                    clusteringChanged = true;
                }
            }
            newClusters.add(new Cluster<T>(newCenter));
        }
        if (!clusteringChanged) {
            return clusters;
        }
        assignPointsToClusters(newClusters, points);
        clusters = newClusters;
    }
    return clusters;
}

---

From source file:org.basketball.MyKMeansPlusPlusClusterer.java

/**
 * Use K-means++ to choose the initial centers.
 *
 * @param <T> type of the points to cluster
 * @param points the points to choose the initial centers from
 * @param k the number of centers to choose
 * @param random random generator to use
 * @return the initial centers//from  w  ww  . jav  a 2 s  . c  o  m
 */
private static <T extends Clusterable<T>> List<Cluster<T>> chooseInitialCenters(final Collection<T> points,
        final int k, final Random random) {

    final List<T> pointSet = new ArrayList<T>(points);
    final List<Cluster<T>> resultSet = new ArrayList<Cluster<T>>();

    // Choose one center uniformly at random from among the data points.
    final T firstPoint = pointSet.remove(random.nextInt(pointSet.size()));
    resultSet.add(new Cluster<T>(firstPoint));

    final double[] dx2 = new double[pointSet.size()];
    while (resultSet.size() < k) {
        // For each data point x, compute D(x), the distance between x and
        // the nearest center that has already been chosen.
        int sum = 0;
        for (int i = 0; i < pointSet.size(); i++) {
            final T p = pointSet.get(i);
            final Cluster<T> nearest = getNearestCluster(resultSet, p);
            final double d = p.distanceFrom(nearest.getCenter());
            sum += d * d;
            dx2[i] = sum;
        }

        // Add one new data point as a center. Each point x is chosen with
        // probability proportional to D(x)2
        final double r = random.nextDouble() * sum;
        for (int i = 0; i < dx2.length; i++) {
            if (dx2[i] >= r) {
                final T p = pointSet.remove(i);
                resultSet.add(new Cluster<T>(p));
                break;
            }
        }
    }

    return resultSet;

}

---