keel.Algorithms.Decision_Trees.SLIQ

Class Node

java.lang.Object

keel.Algorithms.Decision_Trees.SLIQ.Node

public class Node
extends java.lang.Object

This class implements the nodes of SLIQ decision tree.

Version:

1.0 (28/12/09 - 10/1/10)

Author:

Francisco Charte Ojeda

Constructor Summary

Constructors

Constructor and Description
Node(int nClases) Paramenter constructor.

Method Summary

Modifier and Type	Method and Description
void	actualizaClasePrincipal() Actualizes the main class of the node by considering the frecuency of each class.
protected void	actualizaHistograma(int clase) Changes an element with the class given from the left leaf to the right one.
void	addChildren(Node node) Adds a child to the node.
void	agregaElemento(int clase) Adds an element of the class given to the node.
void	calculaCoste(int fase) Computes the cost of each node of the tree.
double	calculaGini() Computes and returns the Gini index for the node (continuous attributes).
double	calculaGini(int indSubconjunto, int[][] ocurrencias, int numValores, int numClases, int totalOcurrencias) Computes and returns the Gini index for the node (discrete attributes).
int	costeError(Node hijo) Computes the error cost of adding a child to the node.
void	divide() Divides the nodes into its two children.
boolean	esHoja() Checks if the node is a leaf.
Node[]	getChildren() Returns the children of the node.
Node	getChildren(int index) Returns the child with the given index.
int	getClase() Returns the class of the node.
int	getCoste() Returns the associated cost.
java.util.Vector<ListaAtributos>[]	getData() Returns the dataset that satisfies the node condition.
int	getDecompositionAttribute() Retuns the dataset that satisfies the node's condition.
double	getDecompositionValue() Returns the value used to divide the node.
double	getIndiceGini() Returns the the Gini index.
Node	getParent() Returns the father node.
int	numChildren() Returns the number of children.
void	pruebaCorte(int indAtributo, ListaClases[] listaClases, Attribute atributo) Checks a given cut and computes a possible improvement.
void	pruebaCorte(int atributo, ListaClases[] listaClases, double valor, double siguiente) Checks a given cut and computes a possible improvement.
void	setChildren(Node[] nodes) Sets the children of the node.
void	setCoste(int coste) Sets the node cost with the value given.
void	setData(java.util.Vector<ListaAtributos>[] newData) Sets the dataset that satisfies the node's condition.
void	setHoja(boolean b) Sets the leaf contition.
void	setParent(Node node) Sets the father of the node with the given node.

Methods inherited from class java.lang.Object

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

Constructor Detail

Node

public Node(int nClases)

Paramenter constructor. The node structures will be initialized with the parameters given.

Parameters:

nClases - number of classes.

Method Detail

agregaElemento

public void agregaElemento(int clase)

Adds an element of the class given to the node.

Parameters:

clase - given class.

divide

public void divide()

Divides the nodes into its two children.

actualizaHistograma

protected void actualizaHistograma(int clase)

Changes an element with the class given from the left leaf to the right one.

Parameters:

clase - class index from the element to be changed.

actualizaClasePrincipal

public void actualizaClasePrincipal()

Actualizes the main class of the node by considering the frecuency of each class. Used after pruning.

pruebaCorte

public void pruebaCorte(int indAtributo,
                        ListaClases[] listaClases,
                        Attribute atributo)

Checks a given cut and computes a possible improvement. (Discrete attributes).

Parameters:

indAtributo - attribute index.

listaClases - classes list.

atributo - attribute reference.

pruebaCorte

public void pruebaCorte(int atributo,
                        ListaClases[] listaClases,
                        double valor,
                        double siguiente)

Checks a given cut and computes a possible improvement. (Continuous attributes).

Parameters:

atributo - attribute index.

listaClases - classes list.

valor - cut value to check.

siguiente - next value.

calculaGini

public double calculaGini()

Computes and returns the Gini index for the node (continuous attributes).

Returns:

the Gini index for the node

calculaGini

public double calculaGini(int indSubconjunto,
                          int[][] ocurrencias,
                          int numValores,
                          int numClases,
                          int totalOcurrencias)

Computes and returns the Gini index for the node (discrete attributes). Based on the implementation of the class Count_Matrix of the Nathan Rountree thesis called 'Initialising Neural Networks with Prior Knowledge'. In particular, the chapter where a study of decision trees and its splitting and pruning methods is done.

Parameters:

indSubconjunto - subset index to check.

ocurrencias - occurrences matrix.

numValores - number of values in the matrix.

numClases - number of classes in the matrix.

totalOcurrencias - total number of occurrences

Returns:

Gini index.

getIndiceGini

public double getIndiceGini()

Returns the the Gini index.

Returns:

the the Gini index.

setData

public void setData(java.util.Vector<ListaAtributos>[] newData)

Sets the dataset that satisfies the node's condition.

Parameters:

newData - Given dataset.

esHoja

public boolean esHoja()

Checks if the node is a leaf.

Returns:

true if the node is a leaf.

setHoja

public void setHoja(boolean b)

Sets the leaf contition.

Parameters:

b - true if the node is a leaf.

getData

public java.util.Vector<ListaAtributos>[] getData()

Returns the dataset that satisfies the node condition.

Returns:

the dataset that satisfies the node condition.

getClase

public int getClase()

Returns the class of the node.

Returns:

the class of the node.

getDecompositionAttribute

public int getDecompositionAttribute()

Retuns the dataset that satisfies the node's condition.

Returns:

the dataset that satisfies the node's condition.

getDecompositionValue

public double getDecompositionValue()

Returns the value used to divide the node.

Returns:

the value used to divide the node.

setChildren

public void setChildren(Node[] nodes)

Sets the children of the node.

Parameters:

nodes - the children of the node.

addChildren

public void addChildren(Node node)

Adds a child to the node.

Parameters:

node - given child.

numChildren

public int numChildren()

Returns the number of children.

Returns:

the number of children.

getChildren

public Node[] getChildren()

Returns the children of the node.

Returns:

the children of the node.

getChildren

public Node getChildren(int index)

Returns the child with the given index.

Parameters:

index - given child index.

Returns:

the child with the given index.

setParent

public void setParent(Node node)

Sets the father of the node with the given node.

Parameters:

node - given father.

getParent

public Node getParent()

Returns the father node.

Returns:

the father node.

getCoste

public int getCoste()

Returns the associated cost.

Returns:

the associated cost.

calculaCoste

public void calculaCoste(int fase)

Computes the cost of each node of the tree.

Parameters:

fase - Pruning phase identifier (1, first; 2, second).

costeError

public int costeError(Node hijo)

Computes the error cost of adding a child to the node.

Parameters:

hijo - child node to add.

Returns:

rerror cost.

setCoste

public void setCoste(int coste)

Sets the node cost with the value given.

Parameters:

coste - given cost.