com.ibm.able.beans.knn
Class AbleNaiveBayes

java.lang.Object
  com.ibm.able.AbleObject
      com.ibm.able.beans.knn.AbleNaiveBayes

All Implemented Interfaces:

AbleBean, AbleDataBufferManager, AbleEventListener, AbleEventListenerManager, AbleEventQueueManager, AbleEventQueueProcessor, AblePropertyChangeManager, AbleSerializable, AbleTranslateTemplateProvider, java.io.Serializable

public class AbleNaiveBayes

extends AbleObject

implements AbleTranslateTemplateProvider, java.io.Serializable

See Also:

Serialized Form

Field Summary

static java.lang.String	defaultName Value assigned to name by default.
double[]	inNum The numeric input buffer.
static java.lang.String[]	MODE_NAMES Labels for the train, test and run modes for GUI use.
double[]	outNum The numeric output buffer.
static java.lang.String	PropertyDiscr
static int	PropertyDiscrId
static java.lang.String	PropertyM
static int	PropertyMId
static int	RUN Application or Run mode output is produced.
static int	TEST Test mode performance is tested (errors are computed).
static int	TRAIN Training mode training data is stored.

Fields inherited from class com.ibm.able.AbleObject

changed, chgSupport, comment, dataFlowEnabled, destBufferConnections, eventQueue, fileName, inputBuffer, listeners, logger, name, outputBuffer, parent, properties, propertyConnectionMgr, sourceBufferConnections, state, stateChgSupport, trace

Constructor Summary

AbleNaiveBayes()
Default constructor.

AbleNaiveBayes(java.lang.String aName)
Construct a naive bayes bean with specified name.

AbleNaiveBayes(java.lang.String aName, int args)

Method Summary

void	calcPosteriorProb() Calculate posterior probability for all the elements of the attribute-value-class table.
double[]	calcPriorClassProb() Calculate prior class probability from training data.
void	calcPriorProb() Calculate prior probability for all the elements of the attribute-value-class table (uniform distribution is assumed).
void	generateTranslateTemplates(AbleFilter inFilt, AbleFilter outFilt, java.util.Vector fields) The next three methods change the default behaviour of the ABLE filters, making them appropriate for Naive Bayes algorithm.
double	getAccuracy() Get the accuracy.
double	getClass(double[] x) Find to which class the current test example is assigned by Naive Bayes algorithm.
double	getCurrentActualClass() Get the current actual classes corresponding to the current test example.
double	getCurrentLearnedClass() Get the current learned classes corresponding to the current test/run example.
double[]	getCurrentTestExample() Get the current test example.
double[]	getCurrentTrainExample() Get the current training example.
int	getDiscretization() Get metric parameter.
AttributeValueClass	getElement(java.util.Vector v, int attr, int val, int cl) Find an Attribute-Value-Class element in a Vector of Attribute-Value-Class elements and return it.
double	getError() Get the error.
double[]	getInNum()
int	getM() Get metric parameter.
int	getMaxProbIndex(double[] prob) Given an array of class probabilities, get the index corresponding to the class with the largest probability value.
int	getNaiveBayesMode() Get the current operating mode of the naive bayes bean.
int	getNumAttributes() Get numAttr parameter.
int	getNumClasses() Get numClasses parameter.
long	getNumCorrectTestExamples() Get the number of correctly classified test examples.
int	getNumInstances(int clasS) Calculates the number of instances in the training data set belonging to a given class based on information stored in the attribute-value-class table.
long	getNumRecords() Get numRecords parameter.
long	getNumTestExamples() Get the number of test examples seen so far.
double[]	getOutNum()
void	init() Get ready to process - init all the bean members.
void	process() Performs the main, synchronous, standard processing function performed by this bean.
void	reset() Reset the naive bayes bean.
java.util.Vector	setAllTable() Constructs the vector containing all the possible attribute-value-class combinations, not just those found in the training data set (some of them may be found in the test data set and need to be initialized to 0).
protected void	setDefaults() Set up the event queue behavior.
void	setDiscretization(int newDiscr) Set discretization parameter.
void	setM(int newM) Set m parameter - weight given to prior.
void	setNaiveBayesMode(int mode) Set the operating mode of the naive bayes bean.
void	setNaiveBayesParameters(int newAttr, int newCl, long newRec) Set the number records, number attributes and number classes for naive bayes bean.

Methods inherited from class com.ibm.able.AbleObject

addAbleEventListener, addDestBufferConnection, addPropertyChangeListener, addPropertyConnection, addSourceBufferConnection, addStateChangeListener, dataChanged, firePropertyChange, flushAbleEventQueue, getAbleEventListeners, getAbleEventProcessingEnabled, getAbleEventQueueSize, getComment, getDestBufferConnections, getFileName, getInputBuffer, getInputBuffer, getInputBufferAsString, getInputBufferAsStringArray, getInputBufferContents, getLogger, getName, getOutputBuffer, getOutputBuffer, getOutputBufferAsString, getOutputBufferAsStringArray, getOutputBufferContents, getParent, getProperties, getProperty, getPropertyChangeListeners, getPropertyConnectionManager, getSleepTime, getSourceBufferConnections, getState, getTraceLogger, handleAbleEvent, hasInputBuffer, hasOutputBuffer, init, isAbleEventPostingEnabled, isAbleEventProcessingEnabled, isChanged, isConnectable, isDataFlowEnabled, isTimerEventProcessingEnabled, notifyAbleEventListeners, process, processAbleEvent, processBufferConnections, processNoEventProcessingEnabledSituation, processTimerEvent, quitAll, quitEnabledEventProcessing, removeAbleEventListener, removeAllAbleEventListeners, removeAllBufferConnections, removeAllConnections, removeAllPropertyConnections, removeDestBufferConnection, removeProperty, removePropertyChangeListener, removePropertyConnection, removeSourceBufferConnection, removeStateChangeListener, restartEnabledEventProcessing, restoreFromFile, restoreFromFile, restoreFromSerializedFile, restoreFromStream, resumeAll, resumeEnabledEventProcessing, saveToFile, saveToFile, setAbleEventProcessingEnabled, setChanged, setComment, setDataFlowEnabled, setFileName, setInputBuffer, setInputBuffer, setLogger, setName, setOutputBuffer, setOutputBuffer, setParent, setProperties, setProperty, setSleepTime, setState, setTimerEventProcessingEnabled, setTraceLogger, sourceConnectionsOK, startEnabledEventProcessing, suspendAll, suspendEnabledEventProcessing

Methods inherited from class java.lang.Object

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

Field Detail

defaultName

public static final java.lang.String defaultName

Value assigned to name by default.

PropertyMId

public static final int PropertyMId

See Also:

Constant Field Values

PropertyM

public static final java.lang.String PropertyM

See Also:

Constant Field Values

PropertyDiscrId

public static final int PropertyDiscrId

See Also:

Constant Field Values

PropertyDiscr

public static final java.lang.String PropertyDiscr

See Also:

Constant Field Values

inNum

public double[] inNum

The numeric input buffer.

outNum

public double[] outNum

The numeric output buffer.

TRAIN

public static final int TRAIN

Training mode training data is stored.

See Also:

Constant Field Values

TEST

public static final int TEST

Test mode performance is tested (errors are computed).

See Also:

Constant Field Values

RUN

public static final int RUN

Application or Run mode output is produced.

See Also:

Constant Field Values

MODE_NAMES

public static final java.lang.String[] MODE_NAMES

Labels for the train, test and run modes for GUI use.

Constructor Detail

AbleNaiveBayes

public AbleNaiveBayes()
               throws AbleException

Default constructor.

AbleNaiveBayes

public AbleNaiveBayes(java.lang.String aName)
               throws AbleException

Construct a naive bayes bean with specified name.

Parameters:

aName - The object name

AbleNaiveBayes

public AbleNaiveBayes(java.lang.String aName,
                      int args)
               throws AbleException

Method Detail

getInNum

public double[] getInNum()

getOutNum

public double[] getOutNum()

setDefaults

protected void setDefaults()
                    throws AbleException

Set up the event queue behavior. No timer processing and no asynch event processing.

Throws:

AbleException

getNaiveBayesMode

public int getNaiveBayesMode()

Get the current operating mode of the naive bayes bean.

Returns:

int.

setNaiveBayesMode

public void setNaiveBayesMode(int mode)

Set the operating mode of the naive bayes bean.

Parameters:

mode - The current operating mode, train/test/run

setNaiveBayesParameters

public void setNaiveBayesParameters(int newAttr,
                                    int newCl,
                                    long newRec)

Set the number records, number attributes and number classes for naive bayes bean.

Parameters:

newAttr - that defines the number of attributes

newCl - that defines the number of classes

newRec - that defines the number of records

setM

public void setM(int newM)
          throws AbleException

Set m parameter - weight given to prior.

Parameters:

newM -

Throws:

AbleException

getM

public int getM()

Get metric parameter.

setDiscretization

public void setDiscretization(int newDiscr)
                       throws AbleException

Set discretization parameter.

Throws:

AbleException

getDiscretization

public int getDiscretization()

Get metric parameter.

getNumRecords

public long getNumRecords()

Get numRecords parameter.

getNumAttributes

public int getNumAttributes()

Get numAttr parameter.

getNumClasses

public int getNumClasses()

Get numClasses parameter.

getNumTestExamples

public long getNumTestExamples()

Get the number of test examples seen so far.

getNumCorrectTestExamples

public long getNumCorrectTestExamples()

Get the number of correctly classified test examples.

getAccuracy

public double getAccuracy()

Get the accuracy.

getError

public double getError()

Get the error.

getCurrentTrainExample

public double[] getCurrentTrainExample()

Get the current training example.

getCurrentTestExample

public double[] getCurrentTestExample()

Get the current test example.

getCurrentLearnedClass

public double getCurrentLearnedClass()

Get the current learned classes corresponding to the current test/run example.

getCurrentActualClass

public double getCurrentActualClass()

Get the current actual classes corresponding to the current test example.

init

public void init()
          throws AbleException

Get ready to process - init all the bean members.

Specified by:

init in interface AbleBean

Overrides:

init in class AbleObject

Throws:

AbleException - If an error occurs.

See Also:

AbleObject.startEnabledEventProcessing()

reset

public void reset()
           throws AbleException

Reset the naive bayes bean.

Specified by:

reset in interface AbleBean

Overrides:

reset in class AbleObject

Throws:

AbleException - If an error occurs.

See Also:

AbleBean.reset()

getMaxProbIndex

public int getMaxProbIndex(double[] prob)

Given an array of class probabilities, get the index corresponding to the class with the largest probability value.

getElement

public AttributeValueClass getElement(java.util.Vector v,
                                      int attr,
                                      int val,
                                      int cl)

Find an Attribute-Value-Class element in a Vector of Attribute-Value-Class elements and return it. Returns null if the searched element is not in the Vector already.

getClass

public double getClass(double[] x)
                throws AbleException

Find to which class the current test example is assigned by Naive Bayes algorithm.

Throws:

AbleException

calcPriorClassProb

public double[] calcPriorClassProb()
                            throws AbleException

Calculate prior class probability from training data. Looks at the parent's data source to find the fields. Can return null if the parent is null.

Throws:

AbleException

setAllTable

public java.util.Vector setAllTable()
                             throws AbleException

Constructs the vector containing all the possible attribute-value-class combinations, not just those found in the training data set (some of them may be found in the test data set and need to be initialized to 0). Looks at the parent's data source to find the fields. Can return null if the parent is null.

Throws:

AbleException

calcPriorProb

public void calcPriorProb()
                   throws AbleException

Calculate prior probability for all the elements of the attribute-value-class table (uniform distribution is assumed).

Throws:

AbleException

getNumInstances

public int getNumInstances(int clasS)

Calculates the number of instances in the training data set belonging to a given class based on information stored in the attribute-value-class table.

calcPosteriorProb

public void calcPosteriorProb()
                       throws AbleException

Calculate posterior probability for all the elements of the attribute-value-class table.

Throws:

AbleException

process

public void process()
             throws AbleException

Description copied from class: AbleObject

Performs the main, synchronous, standard processing function performed by this bean.

This base method implementation provides tracing only.

Specified by:

process in interface AbleBean

Overrides:

process in class AbleObject

Throws:

AbleException

See Also:

AbleObject.inputBuffer, AbleObject.outputBuffer, AbleBean.process()

generateTranslateTemplates

public void generateTranslateTemplates(AbleFilter inFilt,
                                       AbleFilter outFilt,
                                       java.util.Vector fields)
                                throws AbleException

The next three methods change the default behaviour of the ABLE filters, making them appropriate for Naive Bayes algorithm. Thus, for discrete and categorical attributes, the possible values are converted to consecutive numbers starting from 1 till the number of values. For continuous attributes, the values are discretized, so that they take values from 1 to the number of intervals used for discretization. The class attribute is always discrete or categorical.

Specified by:

generateTranslateTemplates in interface AbleTranslateTemplateProvider

Parameters:

inFilt - The AbleFilter used to convert incoming data.

outFilt - The AbleFilter used to convert outgoing data.

fields - A vector of AbleFields from which to derive the translation template.

Returns:

An array containing two AbleFilter objectshe translation template.

Throws:

AbleException