weka.classifiers.trees.m5
Class Rule

java.lang.Object
  weka.classifiers.trees.m5.Rule

All Implemented Interfaces:

java.io.Serializable

public class Rule

extends java.lang.Object

implements java.io.Serializable

Generates a single m5 tree or rule

Version:

$Revision: 1.10 $

Author:

Mark Hall

See Also:

Serialized Form

Field Summary

protected static int	LEFT
private int	m_classIndex the class index
private Instances	m_covered the instances covered by this rule
private double	m_globalAbsDev the absolute deviation of the class for all the instances
private double	m_globalStdDev the standard deviation of the class for all the instances
private Instances	m_instances the instances covered by this rule
private RuleNode[]	m_internalNodes the corresponding internal nodes.
private double	m_minNumInstances The minimum number of instances to allow at a leaf node
private Instances	m_notCovered the instances not covered by this rule
private int	m_numAttributes the number of attributes
private int	m_numCovered the number of instances covered by this rule
private int	m_numInstances the number of instances in the dataset
private boolean	m_regressionTree Make a regression tree instead of a model tree
private int[]	m_relOps the corresponding relational operators (0 = "<=", 1 = ">")
private RuleNode	m_ruleModel the leaf encapsulating the linear model for this rule
private boolean	m_saveInstances Save instances at each node in an M5 tree for visualization purposes.
private boolean	m_smoothPredictions use the original m5 smoothing procedure
private int[]	m_splitAtts the indexes of the attributes used to split on for this rule
private double[]	m_splitVals the corresponding values of the split points
protected RuleNode	m_topOfTree the top of the m5 tree for this rule
private boolean	m_useTree use a pruned m5 tree rather than make a rule
private boolean	m_useUnpruned Build unpruned tree/rule
protected static int	RIGHT

Constructor Summary

Rule()
Constructor declaration

Method Summary

protected static double	absDev(int attr, Instances inst) Returns the absolute deviation value of the supplied attribute index.
void	buildClassifier(Instances data) Generates a single rule or m5 model tree.
double	classifyInstance(Instance instance) Calculates a prediction for an instance using this rule or M5 model tree
RuleNode	getM5RootNode()
double	getMinNumInstances() Get the minimum number of instances to allow at a leaf node
boolean	getRegressionTree() Get the value of regressionTree.
boolean	getSmoothing() Get whether or not smoothing has been turned on
boolean	getUnpruned() Get whether unpruned tree/rules are being generated
boolean	getUseTree() get whether an m5 tree is being used rather than rules
private void	makeRule() Make the single best rule from a pruned m5 model tree
Instances	notCoveredInstances() Get the instances not covered by this rule
private java.lang.String	ruleToString() Return a description of the rule
void	setMinNumInstances(double minNum) Set the minumum number of instances to allow at a leaf node
void	setRegressionTree(boolean newregressionTree) Set the value of regressionTree.
protected void	setSaveInstances(boolean save) Sets whether instances at each node in an M5 tree should be saved for visualization purposes.
void	setSmoothing(boolean s) Smooth predictions
void	setUnpruned(boolean unpruned) Use unpruned tree/rules
void	setUseTree(boolean u) Use an m5 tree rather than generate rules
protected static double	stdDev(int attr, Instances inst) Returns the standard deviation value of the supplied attribute index.
RuleNode	topOfTree() Returns the top of the tree.
java.lang.String	toString() Return a description of the m5 tree or rule
private java.lang.String	treeToString() Return a description of the m5 tree

Methods inherited from class java.lang.Object

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

Field Detail

LEFT

protected static int LEFT

RIGHT

protected static int RIGHT

m_instances

private Instances m_instances

the instances covered by this rule

m_classIndex

private int m_classIndex

the class index

m_numAttributes

private int m_numAttributes

the number of attributes

m_numInstances

private int m_numInstances

the number of instances in the dataset

m_splitAtts

private int[] m_splitAtts

the indexes of the attributes used to split on for this rule

m_splitVals

private double[] m_splitVals

the corresponding values of the split points

m_internalNodes

private RuleNode[] m_internalNodes

the corresponding internal nodes. Used for smoothing rules.

m_relOps

private int[] m_relOps

the corresponding relational operators (0 = "<=", 1 = ">")

m_ruleModel

private RuleNode m_ruleModel

the leaf encapsulating the linear model for this rule

m_topOfTree

protected RuleNode m_topOfTree

the top of the m5 tree for this rule

m_globalStdDev

private double m_globalStdDev

the standard deviation of the class for all the instances

m_globalAbsDev

private double m_globalAbsDev

the absolute deviation of the class for all the instances

m_covered

private Instances m_covered

the instances covered by this rule

m_numCovered

private int m_numCovered

the number of instances covered by this rule

m_notCovered

private Instances m_notCovered

the instances not covered by this rule

m_useTree

private boolean m_useTree

use a pruned m5 tree rather than make a rule

m_smoothPredictions

private boolean m_smoothPredictions

use the original m5 smoothing procedure

m_saveInstances

private boolean m_saveInstances

Save instances at each node in an M5 tree for visualization purposes.

m_regressionTree

private boolean m_regressionTree

Make a regression tree instead of a model tree

m_useUnpruned

private boolean m_useUnpruned

Build unpruned tree/rule

m_minNumInstances

private double m_minNumInstances

The minimum number of instances to allow at a leaf node

Constructor Detail

Rule

public Rule()

Constructor declaration

Method Detail

buildClassifier

public void buildClassifier(Instances data)
                     throws java.lang.Exception

Generates a single rule or m5 model tree.

Parameters:

data - set of instances serving as training data

Throws:

java.lang.Exception - if the rule has not been generated successfully

classifyInstance

public double classifyInstance(Instance instance)
                        throws java.lang.Exception

Calculates a prediction for an instance using this rule or M5 model tree

Returns:

the prediction

Throws:

if - a prediction can't be made.

java.lang.Exception

topOfTree

public RuleNode topOfTree()

Returns the top of the tree.

makeRule

private void makeRule()
               throws java.lang.Exception

Make the single best rule from a pruned m5 model tree

Throws:

if - something goes wrong.

java.lang.Exception

toString

public java.lang.String toString()

Return a description of the m5 tree or rule

Returns:

a description of the m5 tree or rule as a String

treeToString

private java.lang.String treeToString()

Return a description of the m5 tree

Returns:

a description of the m5 tree as a String

ruleToString

private java.lang.String ruleToString()

Return a description of the rule

Returns:

a description of the rule as a String

setUnpruned

public void setUnpruned(boolean unpruned)

Use unpruned tree/rules

Parameters:

unpruned - true if unpruned tree/rules are to be generated

getUnpruned

public boolean getUnpruned()

Get whether unpruned tree/rules are being generated

Returns:

true if unpruned tree/rules are to be generated

setUseTree

public void setUseTree(boolean u)

Use an m5 tree rather than generate rules

Parameters:

u - true if m5 tree is to be used

getUseTree

public boolean getUseTree()

get whether an m5 tree is being used rather than rules

Returns:

true if an m5 tree is being used.

setSmoothing

public void setSmoothing(boolean s)

Smooth predictions

Parameters:

s - true if smoothing is to be used

getSmoothing

public boolean getSmoothing()

Get whether or not smoothing has been turned on

Returns:

true if smoothing is being used

notCoveredInstances

public Instances notCoveredInstances()

Get the instances not covered by this rule

Returns:

the instances not covered

stdDev

protected static final double stdDev(int attr,
                                     Instances inst)

Returns the standard deviation value of the supplied attribute index.

Parameters:

attr - an attribute index

inst - the instances

Returns:

the standard deviation value

absDev

protected static final double absDev(int attr,
                                     Instances inst)

Returns the absolute deviation value of the supplied attribute index.

Parameters:

attr - an attribute index

inst - the instances

Returns:

the absolute deviation value

setSaveInstances

protected void setSaveInstances(boolean save)

Sets whether instances at each node in an M5 tree should be saved for visualization purposes. Default is to save memory.

Parameters:

save - a boolean value

getRegressionTree

public boolean getRegressionTree()

Get the value of regressionTree.

Returns:

Value of regressionTree.

setRegressionTree

public void setRegressionTree(boolean newregressionTree)

Set the value of regressionTree.

Parameters:

newregressionTree - Value to assign to regressionTree.

setMinNumInstances

public void setMinNumInstances(double minNum)

Set the minumum number of instances to allow at a leaf node

Parameters:

minNum - the minimum number of instances

getMinNumInstances

public double getMinNumInstances()

Get the minimum number of instances to allow at a leaf node

Returns:

a double value

getM5RootNode

public RuleNode getM5RootNode()