weka.classifiers.mi.miti

Class TreeNode

java.lang.Object

weka.classifiers.mi.miti.TreeNode

All Implemented Interfaces:

java.io.Serializable

public class TreeNode
extends java.lang.Object
implements java.io.Serializable

Represents a node in the decision tree.

Version:

$Revision: 10369 $

Author:

Luke Bjerring

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
Split	split

Constructor Summary

Constructors

Constructor and Description
TreeNode(TreeNode parent, java.util.ArrayList<weka.core.Instance> instances) Creates node based on given collection of instances and parent node.

Method Summary

Modifier and Type	Method and Description
void	calculateNodeScore(java.util.HashMap<weka.core.Instance,Bag> instanceBags, boolean unbiasedEstimate, int kBEPPConstant, boolean bagCount, double multiplier) Calculates the node score based on the given arguments.
void	deactivateRelatedInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags, java.util.List<java.lang.String> deactivated) Deactives all instances associated with bags that have at least one instance in the current node.
boolean	isLeafNode() Is node a leaf?
boolean	isPositiveLeaf() Is node a positive leaf?
boolean	isPureNegative(java.util.HashMap<weka.core.Instance,Bag> instanceBags) Checks whether all the instances at the node are associated with positive bags.
boolean	isPurePositive(java.util.HashMap<weka.core.Instance,Bag> instanceBags) Checks whether all the instances at the node are associated with negative bags.
TreeNode	left() Returns the left child in the case of a binary split.
void	makeLeafNode(boolean positiveLeaf) Turns the node into a leaf node.
double	nodeScore() Returns the score for this node.
TreeNode[]	nominals() Returns the children in the case of a nominal-attribute split.
TreeNode	parent() Returns the parent.
void	removeDeactivatedInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags) Removes deactivated instances from the node.
java.lang.String	render(int depth, java.util.HashMap<weka.core.Instance,Bag> instanceBags) Recursively renders this node and its branches as a tabbed out tree
TreeNode	right() Returns the right child in the case of a binary split.
void	splitInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags, AlgorithmConfiguration settings, java.util.Random rand, boolean debug) Splits the instances on this node into the best possible child nodes, according to the settings
boolean	trimNegativeBranches() Recursively removes all branches that do not contain a positive leaf.

Methods inherited from class java.lang.Object

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

Field Detail

split

public Split split

Constructor Detail

TreeNode

public TreeNode(TreeNode parent,
                java.util.ArrayList<weka.core.Instance> instances)

Creates node based on given collection of instances and parent node.

Method Detail

nodeScore

public double nodeScore()

Returns the score for this node.

removeDeactivatedInstances

public void removeDeactivatedInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags)

Removes deactivated instances from the node. Does NOT update the node score.

calculateNodeScore

public void calculateNodeScore(java.util.HashMap<weka.core.Instance,Bag> instanceBags,
                               boolean unbiasedEstimate,
                               int kBEPPConstant,
                               boolean bagCount,
                               double multiplier)

Calculates the node score based on the given arguments.

isLeafNode

public boolean isLeafNode()

Is node a leaf?

isPositiveLeaf

public boolean isPositiveLeaf()

Is node a positive leaf?

isPureNegative

public boolean isPureNegative(java.util.HashMap<weka.core.Instance,Bag> instanceBags)

Checks whether all the instances at the node are associated with positive bags.

isPurePositive

public boolean isPurePositive(java.util.HashMap<weka.core.Instance,Bag> instanceBags)

Checks whether all the instances at the node are associated with negative bags.

makeLeafNode

public void makeLeafNode(boolean positiveLeaf)

Turns the node into a leaf node.

parent

public TreeNode parent()

Returns the parent.

left

public TreeNode left()

Returns the left child in the case of a binary split.

right

public TreeNode right()

Returns the right child in the case of a binary split.

nominals

public TreeNode[] nominals()

Returns the children in the case of a nominal-attribute split.

deactivateRelatedInstances

public void deactivateRelatedInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags,
                                       java.util.List<java.lang.String> deactivated)

Deactives all instances associated with bags that have at least one instance in the current node.

splitInstances

public void splitInstances(java.util.HashMap<weka.core.Instance,Bag> instanceBags,
                           AlgorithmConfiguration settings,
                           java.util.Random rand,
                           boolean debug)

Splits the instances on this node into the best possible child nodes, according to the settings

render

public java.lang.String render(int depth,
                               java.util.HashMap<weka.core.Instance,Bag> instanceBags)

Recursively renders this node and its branches as a tabbed out tree

Returns:

a string containing the node and its children, tabbed to the given depth

trimNegativeBranches

public boolean trimNegativeBranches()

Recursively removes all branches that do not contain a positive leaf. Used to create partial tree for MIRI rule learner.

Returns:

true if a positive leaf was encountered during the trim