weka.filters.supervised.attribute

Class ClassConditionalProbabilities

java.lang.Object

weka.filters.Filter

weka.filters.SimpleFilter

weka.filters.SimpleBatchFilter

weka.filters.supervised.attribute.ClassConditionalProbabilities

All Implemented Interfaces:

java.io.Serializable, CapabilitiesHandler, CapabilitiesIgnorer, CommandlineRunnable, OptionHandler, RevisionHandler, WeightedAttributesHandler, WeightedInstancesHandler

public class ClassConditionalProbabilities
extends SimpleBatchFilter
implements WeightedAttributesHandler, WeightedInstancesHandler

Converts the values of nominal and/or numeric attributes into class conditional probabilities. If there are k classes, then k new attributes are created for each of the original ones, giving pr(att val | class k).

Can be useful for converting nominal attributes with a lot of distinct values into something more manageable for learning schemes that can't handle nominal attributes (as opposed to creating binary indicator attributes). For nominal attributes, the user can specify the number values above which an attribute will be converted by this method. Normal distributions are assumed for numeric attributes.

Valid options are:

 -N
  Don't apply this transformation to numeric attributes

 -C
  Don't apply this transformation to nominal attributes

 -min-values <integer>
  Transform nominal attributes with at least this many values.
  -1 means always transform.

 -output-debug-info
  If set, filter is run in debug mode and
  may output additional info to the console

 -do-not-check-capabilities
  If set, filter capabilities are not checked before filter is built
  (use with caution).

-spread-attribute-weight
  When generating binary attributes, spread weight of old
  attribute across new attributes. Do not give each new attribute the old weight.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected NaiveBayes	m_estimator The Naive Bayes classifier to use for class conditional estimation
protected java.util.Map<java.lang.String,Estimator[]>	m_estimatorLookup A lookup of estimators from Naive Bayes
protected boolean	m_excludeNominalAttributes True if nominal attributes are to be excluded from the transformation
protected boolean	m_excludeNumericAttributes True if numeric attributes are to be excluded from the transformation
protected int	m_nominalConversionThreshold Don't convert nominal attributes with fewer than this number of values. -1 means always convert
protected Remove	m_remove Remove filter to use for creating a set of untouched attributes
protected boolean	m_SpreadAttributeWeight Whether to spread attribute weight when creating binary attributes
protected Instances	m_unchanged The attributes from the original data that are untouched by this transformation

Fields inherited from class weka.filters.Filter

m_Debug, m_DoNotCheckCapabilities, m_FirstBatchDone, m_InputRelAtts, m_InputStringAtts, m_NewBatch, m_OutputRelAtts, m_OutputStringAtts

Constructor Summary

Constructors

Constructor and Description
ClassConditionalProbabilities()

Method Summary

Modifier and Type	Method and Description
protected Instance	convertInstance(Instance current) Convert an input instance
protected Instances	determineOutputFormat(Instances inputFormat) Determines the output format based on the input format and returns this.
Capabilities	getCapabilities() Returns the Capabilities of this filter.
NaiveBayes	getEstimator() Get the naive Bayes estimator in use
boolean	getExcludeNominalAttributes() Get whether nominal attributes are to be excluded from the transformation
boolean	getExcludeNumericAttributes() Get whether numeric attributes are being excluded from the transformation
int	getNominalConversionThreshold() Get the minimum number of values a nominal attribute must have in order to be transformed. -1 indicates no minimum (i.e. transform all nominal attributes)
Remove	getRemoveFilter() Get the remove filter in use
java.lang.String	getRevision() Returns the revision string.
boolean	getSpreadAttributeWeight() If true, when generating attributes, spread weight of old attribute across new attributes.
java.lang.String	globalInfo() Global help info for this method
boolean	input(Instance inst) Input an instance for filtering.
static void	main(java.lang.String[] args) Main method for testing this class
protected Instances	process(Instances instances) Processes the given data (may change the provided dataset) and returns the modified version.
void	setEstimator(NaiveBayes nb) Set the naive Bayes estimator to use
void	setExcludeNominalAttributes(boolean e) Set whether nominal attributes are to be excluded from the transformation
void	setExcludeNumericAttributes(boolean e) Set whether numeric attributes are being excluded from the transformation
void	setNominalConversionThreshold(int n) Set the minimum number of values a nominal attribute must have in order to be transformed. -1 indicates no minimum (i.e. transform all nominal attributes)
void	setRemoveFilter(Remove r)
void	setSpreadAttributeWeight(boolean p) If true, when generating attributes, spread weight of old attribute across new attributes.

Methods inherited from class weka.filters.SimpleBatchFilter

allowAccessToFullInputFormat, batchFinished, hasImmediateOutputFormat

Methods inherited from class weka.filters.SimpleFilter

reset, setInputFormat

Methods inherited from class weka.filters.Filter

batchFilterFile, bufferInput, copyValues, copyValues, debugTipText, doNotCheckCapabilitiesTipText, filterFile, flushInput, getCapabilities, getCopyOfInputFormat, getDebug, getDoNotCheckCapabilities, getInputFormat, getOptions, getOutputFormat, initInputLocators, initOutputLocators, inputFormatPeek, isFirstBatchDone, isNewBatch, isOutputFormatDefined, listOptions, makeCopies, makeCopy, mayRemoveInstanceAfterFirstBatchDone, numPendingOutput, output, outputFormatPeek, outputPeek, postExecution, preExecution, push, push, resetQueue, run, runFilter, setDebug, setDoNotCheckCapabilities, setOptions, setOutputFormat, testInputFormat, toString, useFilter, wekaStaticWrapper

Methods inherited from class java.lang.Object

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

Field Detail

m_excludeNumericAttributes

protected boolean m_excludeNumericAttributes

True if numeric attributes are to be excluded from the transformation

m_excludeNominalAttributes

protected boolean m_excludeNominalAttributes

True if nominal attributes are to be excluded from the transformation

m_nominalConversionThreshold

protected int m_nominalConversionThreshold

Don't convert nominal attributes with fewer than this number of values. -1 means always convert

m_estimator

protected NaiveBayes m_estimator

The Naive Bayes classifier to use for class conditional estimation

m_remove

protected Remove m_remove

Remove filter to use for creating a set of untouched attributes

m_unchanged

protected Instances m_unchanged

The attributes from the original data that are untouched by this transformation

m_estimatorLookup

protected java.util.Map<java.lang.String,Estimator[]> m_estimatorLookup

A lookup of estimators from Naive Bayes

m_SpreadAttributeWeight

protected boolean m_SpreadAttributeWeight

Whether to spread attribute weight when creating binary attributes

Constructor Detail

ClassConditionalProbabilities

public ClassConditionalProbabilities()

Method Detail

main

public static void main(java.lang.String[] args)

Main method for testing this class

Parameters:

args - args

globalInfo

public java.lang.String globalInfo()

Global help info for this method

Specified by:

globalInfo in class SimpleFilter

Returns:

the global help info

getExcludeNumericAttributes

@OptionMetadata(displayName="Exclude numeric attributes",
                description="Don\'t apply this transformation to numeric attributes",
                commandLineParamName="N",
                commandLineParamIsFlag=true,
                commandLineParamSynopsis="-N",
                displayOrder=1)
public boolean getExcludeNumericAttributes()

Get whether numeric attributes are being excluded from the transformation

Returns:

true if numeric attributes are to be excluded

setExcludeNumericAttributes

public void setExcludeNumericAttributes(boolean e)

Set whether numeric attributes are being excluded from the transformation

Parameters:

e - true if numeric attributes are to be excluded

getExcludeNominalAttributes

@OptionMetadata(displayName="Exclude nominal attributes",
                description="Don\'t apply this transformation to nominal attributes",
                commandLineParamName="C",
                commandLineParamIsFlag=true,
                commandLineParamSynopsis="-C",
                displayOrder=2)
public boolean getExcludeNominalAttributes()

Get whether nominal attributes are to be excluded from the transformation

Returns:

true if nominal attributes are to be excluded

setExcludeNominalAttributes

public void setExcludeNominalAttributes(boolean e)

Set whether nominal attributes are to be excluded from the transformation

Parameters:

e - true if nominal attributes are to be excluded

setSpreadAttributeWeight

@OptionMetadata(displayName="Spread weight across new attributes",
                description="When generating attributes, spread weight of old\nattribute across new attributes. Do not give each new attribute the old weight.",
                commandLineParamName="spread-attribute-weight",
                commandLineParamIsFlag=true,
                commandLineParamSynopsis="-spread-attribute-weight",
                displayOrder=3)
public void setSpreadAttributeWeight(boolean p)

If true, when generating attributes, spread weight of old attribute across new attributes. Do not give each new attribute the old weight.

Parameters:

p - whether weight is spread

getSpreadAttributeWeight

public boolean getSpreadAttributeWeight()

If true, when generating attributes, spread weight of old attribute across new attributes. Do not give each new attribute the old weight.

Returns:

whether weight is spread

getNominalConversionThreshold

@OptionMetadata(displayName="Nominal conversion threshold",
                description="Transform nominal attributes with at least this many values.\n-1 means always transform.",
                commandLineParamName="min-values",
                commandLineParamSynopsis="-min-values <integer>",
                displayOrder=3)
public int getNominalConversionThreshold()

Get the minimum number of values a nominal attribute must have in order to be transformed. -1 indicates no minimum (i.e. transform all nominal attributes)

Returns:

the number of values of a nominal attribute after which the transformation applies

setNominalConversionThreshold

public void setNominalConversionThreshold(int n)

Set the minimum number of values a nominal attribute must have in order to be transformed. -1 indicates no minimum (i.e. transform all nominal attributes)

Parameters:

n - the number of values of a nominal attribute after which the transformation applies

determineOutputFormat

protected Instances determineOutputFormat(Instances inputFormat)
                                   throws java.lang.Exception

Description copied from class: SimpleFilter

Determines the output format based on the input format and returns this. In case the output format cannot be returned immediately, i.e., immediateOutputFormat() returns false, then this method will be called from batchFinished().

Specified by:

determineOutputFormat in class SimpleFilter

Parameters:

inputFormat - the input format to base the output format on

Returns:

the output format

Throws:

java.lang.Exception - in case the determination goes wrong

See Also:

SimpleFilter.hasImmediateOutputFormat(), Filter.batchFinished()

process

protected Instances process(Instances instances)
                     throws java.lang.Exception

Description copied from class: SimpleFilter

Processes the given data (may change the provided dataset) and returns the modified version. This method is called in batchFinished().

Specified by:

process in class SimpleFilter

Parameters:

instances - the data to process

Returns:

the modified data

Throws:

java.lang.Exception - in case the processing goes wrong

See Also:

Filter.batchFinished()

convertInstance

protected Instance convertInstance(Instance current)
                            throws java.lang.Exception

Convert an input instance

Parameters:

current - the input instance to convert

Returns:

a transformed instance

Throws:

java.lang.Exception - if a problem occurs

input

public boolean input(Instance inst)
              throws java.lang.Exception

Description copied from class: SimpleBatchFilter

Input an instance for filtering. Filter requires all training instances be read before producing output (calling the method batchFinished() makes the data available). If this instance is part of a new batch, m_NewBatch is set to false.

Overrides:

input in class SimpleBatchFilter

Parameters:

inst - the input instance

Returns:

true if the filtered instance may now be collected with output().

Throws:

java.lang.IllegalStateException - if no input structure has been defined

java.lang.Exception - if something goes wrong

See Also:

SimpleBatchFilter.batchFinished()

getCapabilities

public Capabilities getCapabilities()

Returns the Capabilities of this filter.

Specified by:

getCapabilities in interface CapabilitiesHandler

Overrides:

getCapabilities in class Filter

Returns:

the capabilities of this object

See Also:

Capabilities

getRevision

public java.lang.String getRevision()

Returns the revision string.

Specified by:

getRevision in interface RevisionHandler

Overrides:

getRevision in class Filter

Returns:

the revision

getEstimator

@ProgrammaticProperty
public NaiveBayes getEstimator()

Get the naive Bayes estimator in use

Returns:

the naive Bayes estimator

setEstimator

public void setEstimator(NaiveBayes nb)

Set the naive Bayes estimator to use

Parameters:

nb - the naive Bayes estimator to use

getRemoveFilter

@ProgrammaticProperty
public Remove getRemoveFilter()

Get the remove filter in use

Returns:

setRemoveFilter

public void setRemoveFilter(Remove r)