org.apache.commons.math3.ml.clustering.DBSCANClusterer Maven / Gradle / Ivy
Show all versions of cf4j-recsys Show documentation
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.commons.math3.ml.clustering;
import java.util.ArrayList;
import java.util.Collection;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import org.apache.commons.math3.exception.NotPositiveException;
import org.apache.commons.math3.exception.NullArgumentException;
import org.apache.commons.math3.ml.distance.DistanceMeasure;
import org.apache.commons.math3.ml.distance.EuclideanDistance;
import org.apache.commons.math3.util.MathUtils;
/**
* DBSCAN (density-based spatial clustering of applications with noise) algorithm.
*
* The DBSCAN algorithm forms clusters based on the idea of density connectivity, i.e.
* a point p is density connected to another point q, if there exists a chain of
* points pi, with i = 1 .. n and p1 = p and pn = q,
* such that each pair <pi, pi+1> is directly density-reachable.
* A point q is directly density-reachable from point p if it is in the ε-neighborhood
* of this point.
*
* Any point that is not density-reachable from a formed cluster is treated as noise, and
* will thus not be present in the result.
*
* The algorithm requires two parameters:
*
* - eps: the distance that defines the ε-neighborhood of a point
*
- minPoints: the minimum number of density-connected points required to form a cluster
*
*
* @param type of the points to cluster
* @see DBSCAN (wikipedia)
* @see
* A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise
* @since 3.2
*/
public class DBSCANClusterer extends Clusterer {
/** Maximum radius of the neighborhood to be considered. */
private final double eps;
/** Minimum number of points needed for a cluster. */
private final int minPts;
/** Status of a point during the clustering process. */
private enum PointStatus {
/** The point has is considered to be noise. */
NOISE,
/** The point is already part of a cluster. */
PART_OF_CLUSTER
}
/**
* Creates a new instance of a DBSCANClusterer.
*
* The euclidean distance will be used as default distance measure.
*
* @param eps maximum radius of the neighborhood to be considered
* @param minPts minimum number of points needed for a cluster
* @throws NotPositiveException if {@code eps < 0.0} or {@code minPts < 0}
*/
public DBSCANClusterer(final double eps, final int minPts)
throws NotPositiveException {
this(eps, minPts, new EuclideanDistance());
}
/**
* Creates a new instance of a DBSCANClusterer.
*
* @param eps maximum radius of the neighborhood to be considered
* @param minPts minimum number of points needed for a cluster
* @param measure the distance measure to use
* @throws NotPositiveException if {@code eps < 0.0} or {@code minPts < 0}
*/
public DBSCANClusterer(final double eps, final int minPts, final DistanceMeasure measure)
throws NotPositiveException {
super(measure);
if (eps < 0.0d) {
throw new NotPositiveException(eps);
}
if (minPts < 0) {
throw new NotPositiveException(minPts);
}
this.eps = eps;
this.minPts = minPts;
}
/**
* Returns the maximum radius of the neighborhood to be considered.
* @return maximum radius of the neighborhood
*/
public double getEps() {
return eps;
}
/**
* Returns the minimum number of points needed for a cluster.
* @return minimum number of points needed for a cluster
*/
public int getMinPts() {
return minPts;
}
/**
* Performs DBSCAN cluster analysis.
*
* @param points the points to cluster
* @return the list of clusters
* @throws NullArgumentException if the data points are null
*/
@Override
public List> cluster(final Collection points) throws NullArgumentException {
// sanity checks
MathUtils.checkNotNull(points);
final List> clusters = new ArrayList>();
final Map visited = new HashMap();
for (final T point : points) {
if (visited.get(point) != null) {
continue;
}
final List neighbors = getNeighbors(point, points);
if (neighbors.size() >= minPts) {
// DBSCAN does not care about center points
final Cluster cluster = new Cluster();
clusters.add(expandCluster(cluster, point, neighbors, points, visited));
} else {
visited.put(point, PointStatus.NOISE);
}
}
return clusters;
}
/**
* Expands the cluster to include density-reachable items.
*
* @param cluster Cluster to expand
* @param point Point to add to cluster
* @param neighbors List of neighbors
* @param points the data set
* @param visited the set of already visited points
* @return the expanded cluster
*/
private Cluster expandCluster(final Cluster cluster,
final T point,
final List neighbors,
final Collection points,
final Map visited) {
cluster.addPoint(point);
visited.put(point, PointStatus.PART_OF_CLUSTER);
List seeds = new ArrayList(neighbors);
int index = 0;
while (index < seeds.size()) {
final T current = seeds.get(index);
PointStatus pStatus = visited.get(current);
// only check non-visited points
if (pStatus == null) {
final List currentNeighbors = getNeighbors(current, points);
if (currentNeighbors.size() >= minPts) {
seeds = merge(seeds, currentNeighbors);
}
}
if (pStatus != PointStatus.PART_OF_CLUSTER) {
visited.put(current, PointStatus.PART_OF_CLUSTER);
cluster.addPoint(current);
}
index++;
}
return cluster;
}
/**
* Returns a list of density-reachable neighbors of a {@code point}.
*
* @param point the point to look for
* @param points possible neighbors
* @return the List of neighbors
*/
private List getNeighbors(final T point, final Collection points) {
final List neighbors = new ArrayList();
for (final T neighbor : points) {
if (point != neighbor && distance(neighbor, point) <= eps) {
neighbors.add(neighbor);
}
}
return neighbors;
}
/**
* Merges two lists together.
*
* @param one first list
* @param two second list
* @return merged lists
*/
private List merge(final List one, final List two) {
final Set oneSet = new HashSet(one);
for (T item : two) {
if (!oneSet.contains(item)) {
one.add(item);
}
}
return one;
}
}