com.expleague.ml.clustering.impl.GenericNearestNeighborAlgoritm Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of jmll

Various ML methods implemented by myself and my students

There is a newer version: 1.4.9

package com.expleague.ml.clustering.impl;

import com.expleague.commons.math.metrics.Metric;
import com.expleague.ml.clustering.GenericClusterizationAlgorithm;
import org.jetbrains.annotations.NotNull;

import java.util.Arrays;
import java.util.Collection;
import java.util.Collections;
import java.util.HashSet;
import java.util.function.Function;

/**
 * User: terry
 * Date: 16.01.2010
 */
public class GenericNearestNeighborAlgoritm implements GenericClusterizationAlgorithm {
  private final Metric metric;
  private final double acceptanceDistance;
  private final double rejectionDistance;

  public GenericNearestNeighborAlgoritm(final Metric metric, final double acceptanceDistance, final double rejectionDistance) {
    this.metric = metric;
    this.acceptanceDistance = acceptanceDistance;
    this.rejectionDistance = rejectionDistance;
  }

  @NotNull
  @Override
  public Collection> cluster(final Collection dataSet, final Function data2DVector) {
    final Collection> clusters = new HashSet<>();
    for (final X data : dataSet) {
      final V dataVector = data2DVector.apply(data);
      Collection nearestCluster = null;
      double nearestDistance = Double.MAX_VALUE;
      for (final Collection cluster : clusters) {
        for (final X element : cluster) {
          final double candidateDistance = metric.distance(data2DVector.apply(element), dataVector);
          if (candidateDistance < nearestDistance && candidateDistance < acceptanceDistance) {
            nearestDistance = candidateDistance;
            nearestCluster = cluster;
          } else if (candidateDistance > rejectionDistance) break;
        }
      }
      if (nearestCluster == null) {
        clusters.add(new HashSet<>(Collections.singletonList(data)));
      } else {
        nearestCluster.add(data);
      }
    }
    return clusters;
  }
}