de.jstacs.clustering.hierachical

Class Hclust<T>

java.lang.Object

de.jstacs.clustering.hierachical.Hclust<T>

Type Parameters:

T - the type of the elements clustered

public class Hclust<T>
extends Object

Class for clustering a set of elements of the same kind (T) hierarchically using agglomerative clustering with different linkage methods.

Author:

Jan Grau

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Hclust.Linkage The linkage method for clustering

Constructor Summary

Constructors

Constructor and Description
Hclust(DistanceMetric<T> metric, Hclust.Linkage linkage) Creates a new object for clustering using the supplied distance metric and linkage method.

Method Summary

Methods

Modifier and Type	Method and Description
ClusterTree<Integer>	cluster(double[][] distMat, LinkedList<ClusterTree<Integer>> list, int indexOff) Further clusters the supplied cluster trees using the given distance matrix and creating original indexes for the inner node starting at -indexOff-1 in descending order
ClusterTree<T>	cluster(double[][] distMat, T... objects) Clusters the given objects using the supplied distance matrix, which must be in the same order as the elements provided in objects.
ClusterTree<T>	cluster(int indexOff, double[][] distMat, ClusterTree<T>[] leaves) Clusters the given leaf trees using the supplied distance matrix
ClusterTree<T>	cluster(T... objects) Clusters the supplied objects and return the resulting cluster tree.
ClusterTree<T>	createTree(ClusterTree<Integer> intTree, T... objects) Creates a cluster tree given an index tree using the original indexes refering to the indexes of elements in objects.
static <T> T[][]	cutTree(ClusterTree<T> tree, double distance) Cuts the cluster tree at the specified distance and returns the leaf elements grouped by their origin in the sub-trees below the cut
static <T> ClusterTree<T>[]	cutTree(double distance, ClusterTree<T> tree) Cuts the cluster tree at the given distance and returns the sub-trees below the cut.
double	getDistance(double[][] distMat, ClusterTree<Integer> tree, ClusterTree<Integer> tree2) Returns the distance between the two supplied trees using the linkage method of this Hclust object and the given distance matrix.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Hclust

public Hclust(DistanceMetric<T> metric,
      Hclust.Linkage linkage)

Creates a new object for clustering using the supplied distance metric and linkage method.

Parameters:

metric - the distance metric

linkage - the linkage method

Method Detail

cluster

public ClusterTree<T> cluster(T... objects)
                       throws Exception

Clusters the supplied objects and return the resulting cluster tree.

Parameters:

objects - the objects to be clustered

Returns:

the cluster tree

Throws:

Exception - if the distance matrix could not be created using the DistanceMetric of this Hclust object

cluster

public ClusterTree<Integer> cluster(double[][] distMat,
                           LinkedList<ClusterTree<Integer>> list,
                           int indexOff)

Further clusters the supplied cluster trees using the given distance matrix and creating original indexes for the inner node starting at -indexOff-1 in descending order

Parameters:

distMat - the distance matrix

list - the list of previous trees

indexOff - the offset on the original indexes

Returns:

the joint cluster tree

cluster

public ClusterTree<T> cluster(int indexOff,
                     double[][] distMat,
                     ClusterTree<T>[] leaves)

Clusters the given leaf trees using the supplied distance matrix

Parameters:

indexOff - the offset on the original indexes

distMat - the distance matrix

leaves - the leaves

Returns:

the joint cluster tree

See Also:

cluster(int, double[][], ClusterTree[])

cluster

public ClusterTree<T> cluster(double[][] distMat,
                     T... objects)

Clusters the given objects using the supplied distance matrix, which must be in the same order as the elements provided in objects.

Parameters:

distMat - the distance matrix

objects - the objects to the clustered

Returns:

the cluster tree

cutTree

public static <T> T[][] cutTree(ClusterTree<T> tree,
                double distance)

Cuts the cluster tree at the specified distance and returns the leaf elements grouped by their origin in the sub-trees below the cut

Parameters:

tree - the tree

distance - the cut distance

Returns:

the leaf elements

cutTree

public static <T> ClusterTree<T>[] cutTree(double distance,
                           ClusterTree<T> tree)

Cuts the cluster tree at the given distance and returns the sub-trees below the cut.

Parameters:

distance - the distance

tree - the tree

Returns:

the sub-trees

createTree

public ClusterTree<T> createTree(ClusterTree<Integer> intTree,
                        T... objects)

Creates a cluster tree given an index tree using the original indexes refering to the indexes of elements in objects.

Parameters:

intTree - the index tree

objects - the objects filled into the tree instead of the indexes

Returns:

the cluster tree

See Also:

ClusterTree.getIndexTree()

getDistance

public double getDistance(double[][] distMat,
                 ClusterTree<Integer> tree,
                 ClusterTree<Integer> tree2)

Returns the distance between the two supplied trees using the linkage method of this Hclust object and the given distance matrix.

Parameters:

distMat - the distance matrix

tree - the first tree

tree2 - the second tree

Returns:

the distance between the trees