y.algo
Class Groups.Dendrogram

java.lang.Object
  y.algo.Groups.Dendrogram

Enclosing class:

Groups

public static class Groups.Dendrogram
extends java.lang.Object

This class provides the result of hierarchical clustering algorithms by means of a binary tree structure.

The dendrogram is created using the agglomerative strategy (i.e., a bottom-up approach) based on the following steps:

• At the beginning (i.e., at level 0), each node belongs to its own cluster. At this point, each node of the original graph (the one that will be clustered) is mapped to one of the dendrogram nodes that represent a leaf node (i.e., a node with no children).
• At each step, pairs of clusters are merged while moving up to the hierarchy. In this case, a new dendrogram node is created and the corresponding dissimilarity value between the two merged clusters can be obtained using getDissimilarityValue(Node). The dissimilarity values are monotone increasing starting from zero which corresponds to the leaf nodes of the dendrogram.
• The algorithm continues until all nodes belong to the same cluster. At this point, the root of the dendrogram is reached and can be obtained using method getRoot().

The dendrogram is a binary tree that is directed from the root to the leaves. This means that one can iterate the dendrogram starting with the root node, moving on to its children, then moving on to their children and so on.

From each node of the dendrogram, it is possible to retrieve the set of nodes of the original graph that belong to this dendrogram node using getClusterNodes(Node). This method returns an array of NodeLists that contain the nodes of all clusters that are merged when this dendrogram node was created. For example, say that clusters C1 = {v1, v2} and C2 = {v3, v4} are merged, where {v1, v2, v3, v4} are nodes of the original graph. Say that dendrogram node u is created by this merging. Then, the cluster nodes of u will be returned as two NodeLists where the first contains nodes v1, v2 and the second v3, v4.

Only the leaf nodes of the dendrogram are mapped with a node of the original graph. This means that for all other inner nodes of the dendrogram, method getOriginalNode(Node) returns null.

See Also:

Groups.hierarchicalClustering(Graph, Distances, byte), Groups.hierarchicalClustering(Graph, int, NodeMap, Distances, byte), Groups.hierarchicalClustering(Graph, NodeMap, Distances, byte, double)

Method Summary

NodeList	getChildren(Node parent) Returns a NodeList that contains the child nodes of the given parent node.
NodeList[]	getClusterNodes(Node node) Returns an array of NodeLists that contain the nodes of the original graph that are associated with the given dendrogram node.
double	getDissimilarityValue(Node node) Returns the dissimilarity value associated with the given node of the dendrogram.
int	getLevel(Node node) Returns the level of the given node of the dendrogram.
int	getLevelCount() Returns the number of levels this dendrogram has.
Node	getNodeAtLevel(int level) Returns the node of the dendrogram that belongs to the given level index.
int	getNodeCount() Returns the number of nodes in this dendrogram.
Node	getOriginalNode(Node node) Returns the node of original graph that is mapped with the given leaf node of the dendrogram.
Node	getRoot() Returns the root node of the dendrogram.
boolean	isLeaf(Node node) Returns whether or not the given dendrogram node is a leaf node.

Methods inherited from class java.lang.Object

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

Method Detail

getRoot

public Node getRoot()

Returns the root node of the dendrogram.

A dendrogram traversal can be implemented by starting from this node, then moving on to its children using method getChildren(Node), and so on.

Returns:

the root node of the dendrogram

getChildren

public NodeList getChildren(Node parent)

Returns a NodeList that contains the child nodes of the given parent node.

If the given node is a leaf node, the returned list will be empty.

Parameters:

parent - the parent node for which to retrieve the child nodes

Returns:

the list of child nodes of the given parent node

getLevel

public int getLevel(Node node)

Returns the level of the given node of the dendrogram.

This method expects a Node that belongs to the dendrogram representation and not to the original graph.

Parameters:

node - the given dendrogram node

Returns:

the level of the given dendrogram node

getNodeAtLevel

public Node getNodeAtLevel(int level)

Returns the node of the dendrogram that belongs to the given level index.

If the given level index is greater than the number of levels, then this method will return null.

Parameters:

level - the given level index

Returns:

the node of the dendrogram that belongs to the given level index or null if there exists no level with the given index.

Throws:

java.lang.IllegalArgumentException - if the given level is negative or greater than the node count minus one.

getOriginalNode

public Node getOriginalNode(Node node)

Returns the node of original graph that is mapped with the given leaf node of the dendrogram.

This method is meant to be called only for the leaf nodes of the dendrogram and returns null if the given node is not a leaf node.

Parameters:

node - the given leaf node

Returns:

the node of original graph that is mapped with the given leaf node or null if the given node is an inner dendrogram node

getDissimilarityValue

public double getDissimilarityValue(Node node)

Returns the dissimilarity value associated with the given node of the dendrogram.

This method expects a Node that belongs to the dendrogram representation and not to the original graph.

Parameters:

node - the given dendrogram node

Returns:

the dissimilarity value associated with the given dendrogram node

getClusterNodes

public NodeList[] getClusterNodes(Node node)

Returns an array of NodeLists that contain the nodes of the original graph that are associated with the given dendrogram node.

Each such NodeList represents nodes that belong to the same cluster. For example, say that clusters C1 = {v1, v2} and C2 = {v3, v4} are merged, where {v1, v2, v3, v4} are nodes of the original graph. Say that dendrogram node u is created by this merging. Then, the cluster nodes of u will be returned as two NodeLists where the first contains nodes v1, v2 and the second v3, v4.

This method expects a Node that belongs to the dendrogram representation and not to the original graph.

Parameters:

node - the given dendrogram node

Returns:

an array of NodeLists that contain the nodes of the original graph that are associated with the given dendrogram node

isLeaf

public boolean isLeaf(Node node)

Returns whether or not the given dendrogram node is a leaf node.

A leaf node has no further children and represents a node of the original graph. The original node can be retrieved via getOriginalNode(Node).

This method expects a Node that belongs to the dendrogram representation and not to the original graph.

Parameters:

node - a dendrogram node

Returns:

true if the given node is a leaf node, false otherwise

getNodeCount

public int getNodeCount()

Returns the number of nodes in this dendrogram.

Returns:

the number of nodes in this dendrogram

getLevelCount

public int getLevelCount()

Returns the number of levels this dendrogram has.

Returns:

the number of levels of this dendrogram