weka.associations
Class Tertius

java.lang.Object
  weka.associations.Associator
      weka.associations.Tertius

All Implemented Interfaces:

java.lang.Cloneable, OptionHandler, java.lang.Runnable, java.io.Serializable

public class Tertius

extends Associator

implements OptionHandler, java.lang.Runnable

Class implementing a Tertius-type algorithm.

References: P. A. Flach, N. Lachiche (1999). Confirmation-Guided Discovery of first-order rules with Tertius. Machine Learning, 42, 61-95.

Valid options are:

-K number of values in result
Set maximum number of confirmation values in the result. (default: 10)

-F frequency threshold
Set frequency threshold for pruning. (default: 0)

-C confirmation threshold
Set confirmation threshold. (default: 0)

-N noise threshold
Set noise threshold : maximum frequency of counter-examples. 0 gives only satisfied rules. (default: 1)

-R
Allow attributes to be repeated in a same rule.

-L number of literals
Set maximum number of literals in a rule. (default: 4)

-G 0=no negation | 1=body | 2=head | 3=body and head
Set the negations in the rule. (default: 0)

-S
Consider only classification rules.

-c class index
Set index of class attribute. (default: last).

-H
Consider only horn clauses.

-E
Keep equivalent rules.

-M
Keep same clauses.

-T
Keep subsumed rules.

-I 0=always match | 1=never match | 2=significant
Set the way to handle missing values. (default: 0)

-O
Use ROC analysis.

-p name of file
Set the file containing the parts of the individual for individual-based learning.

-P 0=no output | 1=on stdout | 2=in separate window
Set output of current values. (default: 0)

Version:

$Revision: 1.1 $

Author:

Amelie Deltour

See Also:

Serialized Form

Field Summary

private static int	ALL
private static int	BODY
static int	EXPLICIT Ways of handling missing values.
private static int	HEAD
static int	IMPLICIT
private int	m_best Number of best confirmation values to search.
private boolean	m_classification Classification bias.
private int	m_classIndex Index of class attribute.
private double	m_confirmationThreshold Confirmation threshold for the rules.
private boolean	m_equivalent Perform test on equivalent rules ?
private int	m_explored Number of hypotheses explored.
private double	m_frequencyThreshold Frequency threshold for the body and the negation of the head.
private boolean	m_horn Horn clauses bias.
private int	m_hypotheses Number of hypotheses considered.
private Instances	m_instances Instances used for the search.
private int	m_missing Way of handling missing values in the search.
private int	m_negation Type of negation used in the rules.
private double	m_noiseThreshold Maximal number of counter-instances.
private int	m_numLiterals Number of literals in a rule.
private Instances	m_parts Part instances for individual-based learning.
private java.lang.String	m_partsString Name of the file containing the parts for individual-based learning.
private java.util.ArrayList	m_predicates Predicates used in the rules.
private int	m_printValues Type of values output.
private boolean	m_repeat Repeat attributes ?
private SimpleLinkedList	m_results The results.
private boolean	m_roc Perform ROC analysis ?
private boolean	m_sameClause Perform test on same clauses ?
private int	m_status Status of the search.
private boolean	m_subsumption Perform subsumption test ?
private java.util.Date	m_time Time needed for the search.
private java.awt.TextField	m_valuesText Field to output the current values.
private static int	MEMORY
private static int	NO
private static int	NONE Types of negation.
private static int	NORMAL
private static int	OUT
static int	SIGNIFICANT
private static int	STOP
private static Tag[]	TAGS_MISSING
private static Tag[]	TAGS_NEGATION
private static Tag[]	TAGS_VALUES
private static int	WINDOW

Constructor Summary

Tertius()
Constructor that sets the options to the default values.

Method Summary

private void	addResult(Rule rule) Add a rule in the appropriate place in the list of the results, according to the confirmation and number of counter-instances of the rule.
private void	beginSearch() Begin the search by starting a new thread.
void	buildAssociations(Instances instances) Method that launches the search to find the rules with the highest confirmation.
private Predicate	buildPredicate(Instances instances, Attribute attr, boolean isClass) Build the predicate corresponding to an attribute.
private java.util.ArrayList	buildPredicates()
private boolean	canCalculateConfirmation(Rule rule) Test if it is worth calculating the confirmation of a rule.
private boolean	canCalculateOptimistic(Rule rule) Test if it is worth calculating the optimistic estimate of a rule.
private boolean	canExplore(Rule rule) Test if a rule can be explored (if it is interesting for the results or for refining).
private boolean	canRefine(Rule rule) Test if it is worth refining a rule.
private boolean	canStoreInNodes(Rule rule) Test if a rule can be stored in the agenda.
private boolean	canStoreInResults(Rule rule) Test if a rule can be added to the results.
java.lang.String	classificationTipText() Returns the tip text for this property.
java.lang.String	classIndexTipText() Returns the tip text for this property.
java.lang.String	confirmationThresholdTipText() Returns the tip text for this property.
java.lang.String	confirmationValuesTipText() Returns the tip text for this property.
boolean	disabled_getEquivalent() Get the value of equivalent.
java.io.File	disabled_getPartFile() Get the value of partFile.
boolean	disabled_getSameClause() Get the value of sameClause.
boolean	disabled_getSubsumption() Get the value of subsumption.
void	disabled_setEquivalent(boolean v) Set the value of equivalent.
void	disabled_setPartFile(java.io.File v) Set the value of partFile.
void	disabled_setSameClause(boolean v) Set the value of sameClause.
void	disabled_setSubsumption(boolean v) Set the value of subsumption.
private void	endSearch() End the search by notifying to the waiting thread that it is finished.
java.lang.String	equivalentTipText() Returns the tip text for this property.
java.lang.String	frequencyThresholdTipText() Returns the tip text for this property.
boolean	getClassification() Get the value of classification.
int	getClassIndex() Get the value of classIndex.
double	getConfirmationThreshold() Get the value of confirmationThreshold.
int	getConfirmationValues() Get the value of confirmationValues.
double	getFrequencyThreshold() Get the value of frequencyThreshold.
boolean	getHornClauses() Get the value of hornClauses.
SelectedTag	getMissingValues() Get the value of missingValues.
SelectedTag	getNegation() Get the value of negation.
double	getNoiseThreshold() Get the value of noiseThreshold.
int	getNumberLiterals() Get the value of numberLiterals.
java.lang.String[]	getOptions() Gets the current settings of the Tertius object.
boolean	getRepeatLiterals() Get the value of repeatLiterals.
boolean	getRocAnalysis() Get the value of rocAnalysis.
SelectedTag	getValuesOutput() Get the value of valuesOutput.
java.lang.String	globalInfo() Returns a string describing this associator.
java.lang.String	hornClausesTipText() Returns the tip text for this property.
java.util.Enumeration	listOptions() Returns an enumeration describing the available options.
static void	main(java.lang.String[] options) Main method for testing this class.
java.lang.String	missingValuesTipText() Returns the tip text for this property.
java.lang.String	negationTipText() Returns the tip text for this property.
java.lang.String	noiseThresholdTipText() Returns the tip text for this property.
java.lang.String	numberLiteralsTipText() Returns the tip text for this property.
private int	numValuesInResult() Count the number of distinct confirmation values in the results.
java.lang.String	partFileTipText() Returns the tip text for this property.
private void	printValues() Print the current best and worst values.
java.lang.String	repeatLiteralsTipText() Returns the tip text for this property.
void	resetOptions() Resets the options to the default values.
java.lang.String	rocAnalysisTipText() Returns the tip text for this property.
void	run() Run the search.
java.lang.String	sameClauseTipText() Returns the tip text for this property.
void	search() Search in the space of hypotheses the rules that have the highest confirmation.
void	setClassification(boolean v) Set the value of classification.
void	setClassIndex(int v) Set the value of classIndex.
void	setConfirmationThreshold(double v) Set the value of confirmationThreshold.
void	setConfirmationValues(int v) Set the value of confirmationValues.
void	setFrequencyThreshold(double v) Set the value of frequencyThreshold.
void	setHornClauses(boolean v) Set the value of hornClauses.
void	setMissingValues(SelectedTag v) Set the value of missingValues.
void	setNegation(SelectedTag v) Set the value of negation.
void	setNoiseThreshold(double v) Set the value of noiseThreshold.
void	setNumberLiterals(int v) Set the value of numberLiterals.
void	setOptions(java.lang.String[] options) Parses a given list of options.
void	setRepeatLiterals(boolean v) Set the value of repeatLiterals.
void	setRocAnalysis(boolean v) Set the value of rocAnalysis.
void	setValuesOutput(SelectedTag v) Set the value of valuesOutput.
java.lang.String	subsumptionTipText() Returns the tip text for this property.
java.lang.String	toString() Outputs the best rules found with their confirmation value and number of counter-instances.
java.lang.String	valuesOutputTipText() Returns the tip text for this property.

Methods inherited from class weka.associations.Associator

forName, makeCopies

Methods inherited from class java.lang.Object

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

Field Detail

m_results

private SimpleLinkedList m_results

The results.

m_hypotheses

private int m_hypotheses

Number of hypotheses considered.

m_explored

private int m_explored

Number of hypotheses explored.

m_time

private java.util.Date m_time

Time needed for the search.

m_valuesText

private java.awt.TextField m_valuesText

Field to output the current values.

m_instances

private Instances m_instances

Instances used for the search.

m_predicates

private java.util.ArrayList m_predicates

Predicates used in the rules.

m_status

private int m_status

Status of the search.

NORMAL

private static final int NORMAL

See Also:

Constant Field Values

MEMORY

private static final int MEMORY

See Also:

Constant Field Values

STOP

private static final int STOP

See Also:

Constant Field Values

m_best

private int m_best

Number of best confirmation values to search.

m_frequencyThreshold

private double m_frequencyThreshold

Frequency threshold for the body and the negation of the head.

m_confirmationThreshold

private double m_confirmationThreshold

Confirmation threshold for the rules.

m_noiseThreshold

private double m_noiseThreshold

Maximal number of counter-instances.

m_repeat

private boolean m_repeat

Repeat attributes ?

m_numLiterals

private int m_numLiterals

Number of literals in a rule.

NONE

private static final int NONE

Types of negation.

See Also:

Constant Field Values

BODY

private static final int BODY

See Also:

Constant Field Values

HEAD

private static final int HEAD

See Also:

Constant Field Values

ALL

private static final int ALL

See Also:

Constant Field Values

TAGS_NEGATION

private static final Tag[] TAGS_NEGATION

m_negation

private int m_negation

Type of negation used in the rules.

m_classification

private boolean m_classification

Classification bias.

m_classIndex

private int m_classIndex

Index of class attribute.

m_horn

private boolean m_horn

Horn clauses bias.

m_equivalent

private boolean m_equivalent

Perform test on equivalent rules ?

m_sameClause

private boolean m_sameClause

Perform test on same clauses ?

m_subsumption

private boolean m_subsumption

Perform subsumption test ?

EXPLICIT

public static final int EXPLICIT

Ways of handling missing values.

See Also:

Constant Field Values

IMPLICIT

public static final int IMPLICIT

See Also:

Constant Field Values

SIGNIFICANT

public static final int SIGNIFICANT

See Also:

Constant Field Values

TAGS_MISSING

private static final Tag[] TAGS_MISSING

m_missing

private int m_missing

Way of handling missing values in the search.

m_roc

private boolean m_roc

Perform ROC analysis ?

m_partsString

private java.lang.String m_partsString

Name of the file containing the parts for individual-based learning.

m_parts

private Instances m_parts

Part instances for individual-based learning.

NO

private static final int NO

See Also:

Constant Field Values

OUT

private static final int OUT

See Also:

Constant Field Values

WINDOW

private static final int WINDOW

See Also:

Constant Field Values

TAGS_VALUES

private static final Tag[] TAGS_VALUES

m_printValues

private int m_printValues

Type of values output.

Constructor Detail

Tertius

public Tertius()

Constructor that sets the options to the default values.

Method Detail

globalInfo

public java.lang.String globalInfo()

Returns a string describing this associator.

Returns:

A description of the evaluator suitable for displaying in the explorer/experimenter gui.

resetOptions

public void resetOptions()

Resets the options to the default values.

listOptions

public java.util.Enumeration 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 Tertius object.

Specified by:

getOptions in interface OptionHandler

Returns:

An array of strings suitable for passing to setOptions.

confirmationValuesTipText

public java.lang.String confirmationValuesTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getConfirmationValues

public int getConfirmationValues()

Get the value of confirmationValues.

Returns:

Value of confirmationValues.

setConfirmationValues

public void setConfirmationValues(int v)
                           throws java.lang.Exception

Set the value of confirmationValues.

Parameters:

v - Value to assign to confirmationValues.

Throws:

java.lang.Exception

frequencyThresholdTipText

public java.lang.String frequencyThresholdTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getFrequencyThreshold

public double getFrequencyThreshold()

Get the value of frequencyThreshold.

Returns:

Value of frequencyThreshold.

setFrequencyThreshold

public void setFrequencyThreshold(double v)

Set the value of frequencyThreshold.

Parameters:

v - Value to assign to frequencyThreshold.

confirmationThresholdTipText

public java.lang.String confirmationThresholdTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getConfirmationThreshold

public double getConfirmationThreshold()

Get the value of confirmationThreshold.

Returns:

Value of confirmationThreshold.

setConfirmationThreshold

public void setConfirmationThreshold(double v)

Set the value of confirmationThreshold.

Parameters:

v - Value to assign to confirmationThreshold.

noiseThresholdTipText

public java.lang.String noiseThresholdTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getNoiseThreshold

public double getNoiseThreshold()

Get the value of noiseThreshold.

Returns:

Value of noiseThreshold.

setNoiseThreshold

public void setNoiseThreshold(double v)

Set the value of noiseThreshold.

Parameters:

v - Value to assign to noiseThreshold.

repeatLiteralsTipText

public java.lang.String repeatLiteralsTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getRepeatLiterals

public boolean getRepeatLiterals()

Get the value of repeatLiterals.

Returns:

Value of repeatLiterals.

setRepeatLiterals

public void setRepeatLiterals(boolean v)

Set the value of repeatLiterals.

Parameters:

v - Value to assign to repeatLiterals.

numberLiteralsTipText

public java.lang.String numberLiteralsTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getNumberLiterals

public int getNumberLiterals()

Get the value of numberLiterals.

Returns:

Value of numberLiterals.

setNumberLiterals

public void setNumberLiterals(int v)

Set the value of numberLiterals.

Parameters:

v - Value to assign to numberLiterals.

negationTipText

public java.lang.String negationTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getNegation

public SelectedTag getNegation()

Get the value of negation.

Returns:

Value of negation.

setNegation

public void setNegation(SelectedTag v)

Set the value of negation.

Parameters:

v - Value to assign to negation.

classificationTipText

public java.lang.String classificationTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getClassification

public boolean getClassification()

Get the value of classification.

Returns:

Value of classification.

setClassification

public void setClassification(boolean v)

Set the value of classification.

Parameters:

v - Value to assign to classification.

classIndexTipText

public java.lang.String classIndexTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getClassIndex

public int getClassIndex()

Get the value of classIndex.

Returns:

Value of classIndex.

setClassIndex

public void setClassIndex(int v)

Set the value of classIndex.

Parameters:

v - Value to assign to classIndex.

hornClausesTipText

public java.lang.String hornClausesTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getHornClauses

public boolean getHornClauses()

Get the value of hornClauses.

Returns:

Value of hornClauses.

setHornClauses

public void setHornClauses(boolean v)

Set the value of hornClauses.

Parameters:

v - Value to assign to hornClauses.

equivalentTipText

public java.lang.String equivalentTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

disabled_getEquivalent

public boolean disabled_getEquivalent()

Get the value of equivalent.

Returns:

Value of equivalent.

disabled_setEquivalent

public void disabled_setEquivalent(boolean v)

Set the value of equivalent.

Parameters:

v - Value to assign to equivalent.

sameClauseTipText

public java.lang.String sameClauseTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

disabled_getSameClause

public boolean disabled_getSameClause()

Get the value of sameClause.

Returns:

Value of sameClause.

disabled_setSameClause

public void disabled_setSameClause(boolean v)

Set the value of sameClause.

Parameters:

v - Value to assign to sameClause.

subsumptionTipText

public java.lang.String subsumptionTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

disabled_getSubsumption

public boolean disabled_getSubsumption()

Get the value of subsumption.

Returns:

Value of subsumption.

disabled_setSubsumption

public void disabled_setSubsumption(boolean v)

Set the value of subsumption.

Parameters:

v - Value to assign to subsumption.

missingValuesTipText

public java.lang.String missingValuesTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getMissingValues

public SelectedTag getMissingValues()

Get the value of missingValues.

Returns:

Value of missingValues.

setMissingValues

public void setMissingValues(SelectedTag v)

Set the value of missingValues.

Parameters:

v - Value to assign to missingValues.

rocAnalysisTipText

public java.lang.String rocAnalysisTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getRocAnalysis

public boolean getRocAnalysis()

Get the value of rocAnalysis.

Returns:

Value of rocAnalysis.

setRocAnalysis

public void setRocAnalysis(boolean v)

Set the value of rocAnalysis.

Parameters:

v - Value to assign to rocAnalysis.

partFileTipText

public java.lang.String partFileTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

disabled_getPartFile

public java.io.File disabled_getPartFile()

Get the value of partFile.

Returns:

Value of partFile.

disabled_setPartFile

public void disabled_setPartFile(java.io.File v)
                          throws java.lang.Exception

Set the value of partFile.

Parameters:

v - Value to assign to partFile.

Throws:

java.lang.Exception

valuesOutputTipText

public java.lang.String valuesOutputTipText()

Returns the tip text for this property.

Returns:

Tip text for this property suitable for displaying in the explorer/experimenter GUI.

getValuesOutput

public SelectedTag getValuesOutput()

Get the value of valuesOutput.

Returns:

Value of valuesOutput.

setValuesOutput

public void setValuesOutput(SelectedTag v)

Set the value of valuesOutput.

Parameters:

v - Value to assign to valuesOutput.

buildPredicate

private Predicate buildPredicate(Instances instances,
                                 Attribute attr,
                                 boolean isClass)
                          throws java.lang.Exception

Build the predicate corresponding to an attribute.

Parameters:

instances - The instances this predicates comes from.

attr - The attribute this predicate corresponds to.

isClass - Saying if the attribute is the class attribute.

Returns:

The corresponding Predicate.

Throws:

java.lang.Exception - if the predicate could not be build (the attribute is numeric).

buildPredicates

private java.util.ArrayList buildPredicates()
                                     throws java.lang.Exception

Throws:

java.lang.Exception

numValuesInResult

private int numValuesInResult()

Count the number of distinct confirmation values in the results.

Returns:

Number of confirmation values in the results.

canRefine

private boolean canRefine(Rule rule)

Test if it is worth refining a rule.

Parameters:

rule - The rule to consider.

Returns:

True if the rule can be refined, false otherwise.

canCalculateOptimistic

private boolean canCalculateOptimistic(Rule rule)

Test if it is worth calculating the optimistic estimate of a rule.

Parameters:

rule - The rule to consider.

Returns:

True if the optimistic estimate can be calculated, false otherwise.

canExplore

private boolean canExplore(Rule rule)

Test if a rule can be explored (if it is interesting for the results or for refining).

Parameters:

rule - The rule to consider.

Returns:

True if the rule can be explored, false otherwise.

canStoreInNodes

private boolean canStoreInNodes(Rule rule)

Test if a rule can be stored in the agenda.

Parameters:

rule - The rule to consider.

Returns:

True if the rule can be stored, false otherwise.

canCalculateConfirmation

private boolean canCalculateConfirmation(Rule rule)

Test if it is worth calculating the confirmation of a rule.

Parameters:

rule - The rule to consider.

Returns:

True if the confirmation can be calculated, false otherwise.

canStoreInResults

private boolean canStoreInResults(Rule rule)

Test if a rule can be added to the results.

Parameters:

rule - The rule to consider.

Returns:

True if the rule can be stored, false otherwise.

addResult

private void addResult(Rule rule)

Add a rule in the appropriate place in the list of the results, according to the confirmation and number of counter-instances of the rule.

Subsumption tests are performed and the rule may not be added.

Previous results may also be removed because of sumption.

buildAssociations

public void buildAssociations(Instances instances)
                       throws java.lang.Exception

Method that launches the search to find the rules with the highest confirmation.

Specified by:

buildAssociations in class Associator

Parameters:

instances - The instances to be used for generating the rules.

Throws:

java.lang.Exception - if rules can't be built successfully.

run

public void run()

Run the search.

Specified by:

run in interface java.lang.Runnable

beginSearch

private void beginSearch()
                  throws java.lang.Exception

Begin the search by starting a new thread.

Throws:

java.lang.Exception

endSearch

private void endSearch()

End the search by notifying to the waiting thread that it is finished.

search

public void search()

Search in the space of hypotheses the rules that have the highest confirmation. The search starts with the empty rule, other rules are generated by refinement.

printValues

private void printValues()

Print the current best and worst values.

toString

public java.lang.String toString()

Outputs the best rules found with their confirmation value and number of counter-instances. Also gives the number of hypotheses considered and explored, and the time needed.

main

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

Main method for testing this class.