weka.classifiers.meta
Class SubsetEnsemble

java.lang.Object
  weka.classifiers.AbstractClassifier
      weka.classifiers.SingleClassifierEnhancer
          weka.classifiers.RandomizableSingleClassifierEnhancer
              weka.classifiers.meta.SubsetEnsemble

All Implemented Interfaces:

Serializable, Cloneable, weka.classifiers.Classifier, weka.core.CapabilitiesHandler, weka.core.OptionHandler, weka.core.Randomizable, weka.core.RevisionHandler

public class SubsetEnsemble
extends weka.classifiers.RandomizableSingleClassifierEnhancer

Generates an ensemble using the following approach:
- for each attribute apart from class attribute do:
* create new dataset with only this feature and the class attribute
* remove all instances that contain a missing value
* if no instances left in subset, don't build a classifier for this feature
* if at least 1 instance is left in subset, build base classifier with it
If no classifier gets built at all, use ZeroR as backup model, built on the full dataset.
In addition to the default feature for a subset, a number of random features can be added to the subset before the classifier is trained.
At prediction time, the Vote meta-classifier (using the pre-built classifiers) is used to determing the class probabilities or regression value.

Valid options are:

 -num-slots <num>
  Number of execution slots.
  (default: 1 - i.e. no parallelism)

 -combination-rule <AVG|PROD|MAJ|MIN|MAX|MED>
  The combination rule to use
  (default: AVG)

 -num-random <num>
  Number of random features to use in addition.
  (default: 0)

 -S <num>
  Random number seed.
  (default 1)

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

 -W
  Full name of base classifier.
  (default: weka.classifiers.rules.ZeroR)

 Options specific to classifier weka.classifiers.rules.ZeroR:
 

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

Options after -- are passed to the designated classifier.

Version:

$Revision: 4521 $

Author:

fracpete (fracpete at waikato dot ac dot nz)

See Also:

Serialized Form

Field Summary

protected weka.classifiers.rules.ZeroR	m_BackupModel The backup classifier, in case no ensemble could be constructed at prediction time.
protected weka.classifiers.Classifier[]	m_Classifiers the actual classifiers in use.
protected int	m_CombinationRule Combination Rule variable.
protected int	m_Completed The number of classifiers completed so far
protected weka.core.Instances	m_Data For holding the original training set temporarily.
protected ThreadPoolExecutor	m_ExecutorPool Pool of threads to train models with
protected int	m_Failed The number of classifiers that experienced a failure of some sort during construction.
protected weka.core.Instances	m_Header The header of the training set.
protected int	m_NumExecutionSlots The number of threads to have executing at any one time
protected int	m_NumRandomFeatures the number of random features to use (in addition to base attribute).

Fields inherited from class weka.classifiers.RandomizableSingleClassifierEnhancer

m_Seed

Fields inherited from class weka.classifiers.SingleClassifierEnhancer

m_Classifier

Fields inherited from class weka.classifiers.AbstractClassifier

m_Debug

Constructor Summary

SubsetEnsemble()

Method Summary

void	buildClassifier(weka.core.Instances data) Stump method for building the classifiers
protected void	buildClassifiers() Does the actual construction of the ensemble.
double	classifyInstance(weka.core.Instance instance) Classifies the given test instance.
String	combinationRuleTipText() Returns the tip text for this property.
protected void	completedClassifier(int index, boolean success) Records the completion of the training of a single classifier.
protected weka.classifiers.Classifier	constructEnsemble(weka.core.Instance instance) Constructs the ensemble.
double[]	distributionForInstance(weka.core.Instance instance) Predicts the class memberships for a given instance.
protected int	getActualIndex(int index) Returns the actual index in the data of the feature attribute.
weka.core.SelectedTag	getCombinationRule() Gets the combination rule used
protected weka.filters.Filter	getFilter(int index, int seed, boolean withMissing) Gets a filter for a particular index.
int	getNumExecutionSlots() Get the number of execution slots (threads) to use for building the members of the ensemble.
int	getNumRandomFeatures() Returns the number of additional random features to use.
String[]	getOptions() Gets the current settings of the classifier.
String	getRevision() Returns the revision string.
protected weka.core.Instances	getTrainingSet(int index, int seed) Gets a training set for a particular index.
String	globalInfo() Returns a string describing the classifier.
Enumeration	listOptions() Returns an enumeration describing the available options.
static void	main(String[] args) Main method for running this class from commandline.
String	numExecutionSlotsTipText() Returns the tip text for this property.
String	numRandomFeaturesTipText() Returns the tip text for this property.
void	setCombinationRule(weka.core.SelectedTag value) Sets the combination rule to use.
void	setNumExecutionSlots(int value) Set the number of execution slots (threads) to use for building the members of the ensemble.
void	setNumRandomFeatures(int value) Set the number of additional random features to use.
void	setOptions(String[] options) Parses a given list of options.
protected void	startExecutorPool() Start the pool of execution threads.
String	toString() Returns a string representation of the classifier.

Methods inherited from class weka.classifiers.RandomizableSingleClassifierEnhancer

getSeed, seedTipText, setSeed

Methods inherited from class weka.classifiers.SingleClassifierEnhancer

classifierTipText, defaultClassifierString, getCapabilities, getClassifier, getClassifierSpec, setClassifier

Methods inherited from class weka.classifiers.AbstractClassifier

debugTipText, forName, getDebug, makeCopies, makeCopy, runClassifier, setDebug

Methods inherited from class java.lang.Object

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

Field Detail

m_Classifiers

protected weka.classifiers.Classifier[] m_Classifiers

the actual classifiers in use.

m_NumExecutionSlots

protected int m_NumExecutionSlots

The number of threads to have executing at any one time

m_CombinationRule

protected int m_CombinationRule

Combination Rule variable.

m_NumRandomFeatures

protected int m_NumRandomFeatures

the number of random features to use (in addition to base attribute).

m_ExecutorPool

protected transient ThreadPoolExecutor m_ExecutorPool

Pool of threads to train models with

m_Completed

protected int m_Completed

The number of classifiers completed so far

m_Failed

protected int m_Failed

The number of classifiers that experienced a failure of some sort during construction.

m_Data

protected weka.core.Instances m_Data

For holding the original training set temporarily.

m_Header

protected weka.core.Instances m_Header

The header of the training set.

m_BackupModel

protected weka.classifiers.rules.ZeroR m_BackupModel

The backup classifier, in case no ensemble could be constructed at prediction time.

Constructor Detail

SubsetEnsemble

public SubsetEnsemble()

Method Detail

globalInfo

public String globalInfo()

Returns a string describing the classifier.

Returns:

a description suitable for displaying in the gui

listOptions

public Enumeration listOptions()

Returns an enumeration describing the available options.

Specified by:

listOptions in interface weka.core.OptionHandler

Overrides:

listOptions in class weka.classifiers.RandomizableSingleClassifierEnhancer

Returns:

an enumeration of all the available options.

setOptions

public void setOptions(String[] options)
                throws Exception

Parses a given list of options.

Valid options are:

 -num-slots <num>
  Number of execution slots.
  (default: 1 - i.e. no parallelism)

 -combination-rule <AVG|PROD|MAJ|MIN|MAX|MED>
  The combination rule to use
  (default: AVG)

 -num-random <num>
  Number of random features to use in addition.
  (default: 0)

 -S <num>
  Random number seed.
  (default 1)

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

 -W
  Full name of base classifier.
  (default: weka.classifiers.rules.ZeroR)

 Options specific to classifier weka.classifiers.rules.ZeroR:
 

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

Options after -- are passed to the designated classifier.

Specified by:

setOptions in interface weka.core.OptionHandler

Overrides:

setOptions in class weka.classifiers.RandomizableSingleClassifierEnhancer

Parameters:

options - the list of options as an array of strings

Throws:

Exception - if an option is not supported

getOptions

public String[] getOptions()

Gets the current settings of the classifier.

Specified by:

getOptions in interface weka.core.OptionHandler

Overrides:

getOptions in class weka.classifiers.RandomizableSingleClassifierEnhancer

Returns:

an array of strings suitable for passing to setOptions

setNumExecutionSlots

public void setNumExecutionSlots(int value)

Set the number of execution slots (threads) to use for building the members of the ensemble.

Parameters:

value - the number of slots to use.

getNumExecutionSlots

public int getNumExecutionSlots()

Get the number of execution slots (threads) to use for building the members of the ensemble.

Returns:

the number of slots to use

numExecutionSlotsTipText

public String numExecutionSlotsTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

setCombinationRule

public void setCombinationRule(weka.core.SelectedTag value)

Sets the combination rule to use. Values other than

Parameters:

value - the combination rule method to use

getCombinationRule

public weka.core.SelectedTag getCombinationRule()

Gets the combination rule used

Returns:

the combination rule used

combinationRuleTipText

public String combinationRuleTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

setNumRandomFeatures

public void setNumRandomFeatures(int value)

Set the number of additional random features to use.

Parameters:

value - the number of random features

getNumRandomFeatures

public int getNumRandomFeatures()

Returns the number of additional random features to use.

Returns:

the number of random features

numRandomFeaturesTipText

public String numRandomFeaturesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

startExecutorPool

protected void startExecutorPool()

Start the pool of execution threads.

buildClassifiers

protected void buildClassifiers()
                         throws Exception

Does the actual construction of the ensemble.

Throws:

Exception - if something goes wrong during the training process

completedClassifier

protected void completedClassifier(int index,
                                   boolean success)

Records the completion of the training of a single classifier. Unblocks if all classifiers have been trained.

Parameters:

index - the index of the classifier that has completed

success - whether the classifier trained successfully

getActualIndex

protected int getActualIndex(int index)
                      throws Exception

Returns the actual index in the data of the feature attribute.

Parameters:

index - the index for the requested attribute

Returns:

the actual attribute index for the supplied index

Throws:

Exception - if something goes wrong

getFilter

protected weka.filters.Filter getFilter(int index,
                                        int seed,
                                        boolean withMissing)
                                 throws Exception

Gets a filter for a particular index.

Parameters:

index - the index for the requested filter

seed - the seed value to use for the determining the additional random features

withMissing - whether to include the RemoveInstancesWithMissingValue filter

Returns:

the filter for the supplied index

Throws:

Exception - if something goes wrong

getTrainingSet

protected weka.core.Instances getTrainingSet(int index,
                                             int seed)
                                      throws Exception

Gets a training set for a particular index.

Parameters:

index - the index for the requested training set

seed - the seed value to use for the determining the additional random features

Returns:

the training set for the supplied index

Throws:

Exception - if something goes wrong

buildClassifier

public void buildClassifier(weka.core.Instances data)
                     throws Exception

Stump method for building the classifiers

Parameters:

data - the training data to be used for generating the ensemble

Throws:

Exception - if the classifier could not be built successfully

constructEnsemble

protected weka.classifiers.Classifier constructEnsemble(weka.core.Instance instance)

Constructs the ensemble.

Parameters:

instance - the instance to base the construction on

classifyInstance

public double classifyInstance(weka.core.Instance instance)
                        throws Exception

Classifies the given test instance. The instance has to belong to a dataset when it's being classified.

Specified by:

classifyInstance in interface weka.classifiers.Classifier

Overrides:

classifyInstance in class weka.classifiers.AbstractClassifier

Parameters:

instance - the instance to be classified

Returns:

the predicted most likely class for the instance or Utils.missingValue() if no prediction is made

Throws:

Exception - if an error occurred during the prediction

distributionForInstance

public double[] distributionForInstance(weka.core.Instance instance)
                                 throws Exception

Predicts the class memberships for a given instance. If an instance is unclassified, the returned array elements must be all zero. If the class is numeric, the array must consist of only one element, which contains the predicted value.

Specified by:

distributionForInstance in interface weka.classifiers.Classifier

Overrides:

distributionForInstance in class weka.classifiers.AbstractClassifier

Parameters:

instance - the instance to be classified

Returns:

an array containing the estimated membership probabilities of the test instance in each class or the numeric prediction

Throws:

Exception - if distribution could not be computed successfully

toString

public String toString()

Returns a string representation of the classifier.

Overrides:

toString in class Object

Returns:

the string representation

getRevision

public String getRevision()

Returns the revision string.

Specified by:

getRevision in interface weka.core.RevisionHandler

Overrides:

getRevision in class weka.classifiers.AbstractClassifier

Returns:

the revision

main

public static void main(String[] args)

Main method for running this class from commandline.

Parameters:

args - the options