weka.classifiers.bayes.net.search

Class SearchAlgorithm

java.lang.Object

weka.classifiers.bayes.net.search.SearchAlgorithm

All Implemented Interfaces:

java.io.Serializable, OptionHandler, RevisionHandler

Direct Known Subclasses:

FromFile, GlobalScoreSearchAlgorithm, LocalScoreSearchAlgorithm, NaiveBayes

public class SearchAlgorithm
extends java.lang.Object
implements OptionHandler, java.io.Serializable, RevisionHandler

This is the base class for all search algorithms for learning Bayes networks. It contains some common code, used by other network structure search algorithms, and should not be used by itself.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	m_bInitAsNaiveBayes determines whether initial structure is an empty graph or a Naive Bayes network
protected boolean	m_bMarkovBlanketClassifier Determines whether after structure is found a MarkovBlanketClassifier correction should be applied If this is true, m_bInitAsNaiveBayes is overridden and interpreted as false.
protected int	m_nMaxNrOfParents Holds upper bound on number of parents
protected java.lang.String	m_sInitalBIFFile File name containing initial network structure.

Constructor Summary

Constructors

Constructor and Description
SearchAlgorithm() c'tor

Method Summary

Modifier and Type	Method and Description
protected boolean	addArcMakesSense(BayesNet bayesNet, Instances instances, int iAttributeHead, int iAttributeTail) AddArcMakesSense checks whether adding the arc from iAttributeTail to iAttributeHead does not already exists and does not introduce a cycle
void	buildStructure(BayesNet bayesNet, Instances instances) buildStructure determines the network structure/graph of the network.
protected void	doMarkovBlanketCorrection(BayesNet bayesNet, Instances instances) for each node in the network make sure it is in the Markov blanket of the classifier node, and if not, add arrows so that it is.
protected boolean	getMarkovBlanketClassifier()
java.lang.String[]	getOptions() Gets the current settings of the Classifier.
java.lang.String	getRevision() Returns the revision string.
java.lang.String	initAsNaiveBayesTipText()
protected boolean	isArc(BayesNet bayesNet, int iAttributeHead, int iAttributeTail) IsArc checks whether the arc from iAttributeTail to iAttributeHead already exists
java.util.Enumeration<Option>	listOptions() Returns an enumeration describing the available options.
protected java.lang.String	markovBlanketClassifierTipText()
java.lang.String	maxNrOfParentsTipText()
protected boolean	reverseArcMakesSense(BayesNet bayesNet, Instances instances, int iAttributeHead, int iAttributeTail) reverseArcMakesSense checks whether the arc from iAttributeTail to iAttributeHead exists and reversing does not introduce a cycle
protected void	search(BayesNet bayesNet, Instances instances)
protected void	setMarkovBlanketClassifier(boolean bMarkovBlanketClassifier)
void	setOptions(java.lang.String[] options) Parses a given list of options.
java.lang.String	toString() a string representation of the algorithm

Methods inherited from class java.lang.Object

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

Field Detail

m_nMaxNrOfParents

protected int m_nMaxNrOfParents

Holds upper bound on number of parents

m_bInitAsNaiveBayes

protected boolean m_bInitAsNaiveBayes

determines whether initial structure is an empty graph or a Naive Bayes network

m_bMarkovBlanketClassifier

protected boolean m_bMarkovBlanketClassifier

Determines whether after structure is found a MarkovBlanketClassifier correction should be applied If this is true, m_bInitAsNaiveBayes is overridden and interpreted as false.

m_sInitalBIFFile

protected java.lang.String m_sInitalBIFFile

File name containing initial network structure. This can be used as starting point for structure search It will be ignored if not speficied. When specified, it overrides the InitAsNaivBayes flag.

Constructor Detail

SearchAlgorithm

public SearchAlgorithm()

c'tor

Method Detail

addArcMakesSense

protected boolean addArcMakesSense(BayesNet bayesNet,
                                   Instances instances,
                                   int iAttributeHead,
                                   int iAttributeTail)
                            throws java.lang.InterruptedException

AddArcMakesSense checks whether adding the arc from iAttributeTail to iAttributeHead does not already exists and does not introduce a cycle

Parameters:

bayesNet -

instances -

iAttributeHead - index of the attribute that becomes head of the arrow

iAttributeTail - index of the attribute that becomes tail of the arrow

Returns:

true if adding arc is allowed, otherwise false

Throws:

java.lang.InterruptedException

reverseArcMakesSense

protected boolean reverseArcMakesSense(BayesNet bayesNet,
                                       Instances instances,
                                       int iAttributeHead,
                                       int iAttributeTail)

reverseArcMakesSense checks whether the arc from iAttributeTail to iAttributeHead exists and reversing does not introduce a cycle

Parameters:

bayesNet -

instances -

iAttributeHead - index of the attribute that is head of the arrow

iAttributeTail - index of the attribute that is tail of the arrow

Returns:

true if the arc from iAttributeTail to iAttributeHead exists and reversing does not introduce a cycle

isArc

protected boolean isArc(BayesNet bayesNet,
                        int iAttributeHead,
                        int iAttributeTail)

IsArc checks whether the arc from iAttributeTail to iAttributeHead already exists

Parameters:

bayesNet -

iAttributeHead - index of the attribute that becomes head of the arrow

iAttributeTail - index of the attribute that becomes tail of the arrow

Returns:

true if the arc from iAttributeTail to iAttributeHead already exists

listOptions

public java.util.Enumeration<Option> listOptions()

Returns an enumeration describing the available options.

Specified by:

listOptions in interface OptionHandler

Returns:

an enumeration of all the available options.

setOptions

public void setOptions(java.lang.String[] options)
                throws java.lang.Exception

Parses a given list of options.

Specified by:

setOptions in interface OptionHandler

Parameters:

options - the list of options as an array of strings

Throws:

java.lang.Exception - if an option is not supported

getOptions

public java.lang.String[] getOptions()

Gets the current settings of the Classifier.

Specified by:

getOptions in interface OptionHandler

Returns:

an array of strings suitable for passing to setOptions

toString

public java.lang.String toString()

a string representation of the algorithm

Overrides:

toString in class java.lang.Object

Returns:

a string representation

buildStructure

public void buildStructure(BayesNet bayesNet,
                           Instances instances)
                    throws java.lang.Exception

buildStructure determines the network structure/graph of the network. The default behavior is creating a network where all nodes have the first node as its parent (i.e., a BayesNet that behaves like a naive Bayes classifier). This method can be overridden by derived classes to restrict the class of network structures that are acceptable.

Parameters:

bayesNet - the network

instances - the data to use

Throws:

java.lang.Exception - if something goes wrong

search

protected void search(BayesNet bayesNet,
                      Instances instances)
               throws java.lang.Exception

Parameters:

bayesNet -

instances -

Throws:

java.lang.Exception

doMarkovBlanketCorrection

protected void doMarkovBlanketCorrection(BayesNet bayesNet,
                                         Instances instances)

for each node in the network make sure it is in the Markov blanket of the classifier node, and if not, add arrows so that it is. If the node is an ancestor of the classifier node, add arrow pointing to the classifier node, otherwise, add arrow pointing to attribute node.

Parameters:

bayesNet -

instances -

setMarkovBlanketClassifier

protected void setMarkovBlanketClassifier(boolean bMarkovBlanketClassifier)

Parameters:

bMarkovBlanketClassifier -

getMarkovBlanketClassifier

protected boolean getMarkovBlanketClassifier()

Returns:

maxNrOfParentsTipText

public java.lang.String maxNrOfParentsTipText()

Returns:

a string to describe the MaxNrOfParentsoption.

initAsNaiveBayesTipText

public java.lang.String initAsNaiveBayesTipText()

Returns:

a string to describe the InitAsNaiveBayes option.

markovBlanketClassifierTipText

protected java.lang.String markovBlanketClassifierTipText()

Returns:

a string to describe the MarkovBlanketClassifier option.

getRevision

public java.lang.String getRevision()

Returns the revision string.

Specified by:

getRevision in interface RevisionHandler

Returns:

the revision