Package adams.flow.transformer

Class WekaClassifierRanker.RankingJob

java.lang.Object

adams.core.logging.LoggingObject

adams.core.logging.CustomLoggingLevelObject

adams.multiprocess.AbstractJob

adams.flow.transformer.WekaClassifierRanker.RankingJob

All Implemented Interfaces:

CleanUpHandler, LoggingLevelHandler, LoggingSupporter, SizeOfHandler, Stoppable, StoppableWithFeedback, Job, Serializable

Enclosing class:

WekaClassifierRanker

public static class WekaClassifierRanker.RankingJob
extends AbstractJob

A job class specific to ranking classifiers.

Author:

fracpete (fracpete at waikato dot ac dot nz)

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected weka.classifiers.Classifier	m_BestClassifier	the best classifier.
protected weka.classifiers.Classifier	m_Classifier	the classifier to evaluate.
protected WekaLabelIndex	m_ClassLabel	the class label index.
protected StoppableEvaluation	m_Evaluation	the current evaluation.
protected String	m_EvaluationError	for storing evaluating errors.
protected int	m_Folds	the number of folds to use.
protected int	m_Index	the index in actor's input array.
protected WekaClassifierRanker.Measure	m_Measure	the measure to use for ranking.
protected boolean	m_OutputBestSetup	whether to output the best classifier.
protected weka.classifiers.meta.multisearch.Performance	m_Performance	the performance.
protected long	m_Seed	the seed value to use.
protected weka.core.Instances	m_Test	the test data to evaluate with (if folds less than 2).
protected weka.core.Instances	m_Train	the train data to evaluate with.

Fields inherited from class adams.multiprocess.AbstractJob

m_Complete, m_ExecutionError, m_JobCompleteListener, m_JobInfo, m_ProgressInfo, m_Stopped

Fields inherited from class adams.core.logging.LoggingObject

m_Logger, m_LoggingIsEnabled, m_LoggingLevel

Constructor Summary

Constructors

Constructor	Description
RankingJob(weka.classifiers.Classifier cls, int index, weka.core.Instances train, weka.core.Instances test, long seed, int folds, WekaClassifierRanker.Measure measure, WekaLabelIndex classLabel, boolean best)	Initializes the job.

Method Summary

Modifier and Type	Method	Description
void	cleanUp()	Cleans up data structures, frees up memory.
protected String	getAdditionalErrorInformation()	Returns additional information to be added to the error message.
weka.classifiers.Classifier	getBestClassifier()	Returns the best classifier found.
protected weka.classifiers.Classifier	getBestClassifier(weka.classifiers.Classifier template, weka.classifiers.Classifier trained)	In case of GridSearch/MultiSearch the best setup is returned, otherwise the classifier itself.
weka.classifiers.Classifier	getClassifier()	Returns the classifier being used.
int	getFolds()	Returns the number of folds.
int	getIndex()	Returns the index of the classifier in the actor's input array.
WekaClassifierRanker.Measure	getMeasure()	Returns the measure for ranking.
boolean	getOutputBestSetup()	Returns whether the best setup is output in case of optimizers like GridSearch/MultiSearch.
weka.classifiers.meta.multisearch.Performance	getPerformance()	The generated performance.
long	getSeed()	Returns the seed value.
weka.core.Instances	getTest()	The test data.
weka.core.Instances	getTrain()	The training data.
protected String	postProcessCheck()	Checks whether all post-conditions have been met.
protected String	preProcessCheck()	Checks whether all pre-conditions have been met.
protected void	process()	Does the actual execution of the job.
void	stopExecution()	Stops the execution.
String	toString()	Returns a string representation of the job.

Methods inherited from class adams.multiprocess.AbstractJob

execute, getExecutionError, getJobCompleteListener, getJobInfo, getProgressInfo, hasExecutionError, isComplete, isStopped, jobCompleted, setJobCompleteListener, setJobInfo, setProgressInfo

Methods inherited from class adams.core.logging.CustomLoggingLevelObject

setLoggingLevel

Methods inherited from class adams.core.logging.LoggingObject

configureLogger, getLogger, getLoggingLevel, initializeLogging, isLoggingEnabled, sizeOf

Methods inherited from class java.lang.Object

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

Methods inherited from interface adams.core.logging.LoggingLevelHandler

getLoggingLevel

Methods inherited from interface adams.core.logging.LoggingSupporter

getLogger, isLoggingEnabled

Field Detail

m_Classifier

protected weka.classifiers.Classifier m_Classifier

the classifier to evaluate.

m_Index

protected int m_Index

the index in actor's input array.

m_Train

protected weka.core.Instances m_Train

the train data to evaluate with.

m_Test

protected weka.core.Instances m_Test

the test data to evaluate with (if folds less than 2).

m_Seed

protected long m_Seed

the seed value to use.

m_Folds

protected int m_Folds

the number of folds to use.

m_Measure

protected WekaClassifierRanker.Measure m_Measure

the measure to use for ranking.

m_ClassLabel

protected WekaLabelIndex m_ClassLabel

the class label index.

m_Performance

protected weka.classifiers.meta.multisearch.Performance m_Performance

the performance.

m_EvaluationError

protected String m_EvaluationError

for storing evaluating errors.

m_OutputBestSetup

protected boolean m_OutputBestSetup

whether to output the best classifier.

m_BestClassifier

protected weka.classifiers.Classifier m_BestClassifier

the best classifier.

m_Evaluation

protected transient StoppableEvaluation m_Evaluation

the current evaluation.

Constructor Detail

RankingJob

public RankingJob(weka.classifiers.Classifier cls,
                  int index,
                  weka.core.Instances train,
                  weka.core.Instances test,
                  long seed,
                  int folds,
                  WekaClassifierRanker.Measure measure,
                  WekaLabelIndex classLabel,
                  boolean best)

Initializes the job.

Parameters:

cls - the classifier to evaluate

index - the index of the classifier in input array

train - the training data to use

test - the test data to use

seed - the seed value to use

folds - the number of folds to use

measure - the measure to use for ranking

classLabel - the class label index to use

best - whether to output the best classifier setup (for optimizers)

Method Detail

getClassifier

public weka.classifiers.Classifier getClassifier()

Returns the classifier being used.

Returns:

the classifier in use

getIndex

public int getIndex()

Returns the index of the classifier in the actor's input array.

Returns:

the index

getTrain

public weka.core.Instances getTrain()

The training data.

Returns:

the training data

getTest

public weka.core.Instances getTest()

The test data.

Returns:

the test data

getSeed

public long getSeed()

Returns the seed value.

Returns:

the seed value

getFolds

public int getFolds()

Returns the number of folds.

Returns:

the folds

getMeasure

public WekaClassifierRanker.Measure getMeasure()

Returns the measure for ranking.

Returns:

the measure

getPerformance

public weka.classifiers.meta.multisearch.Performance getPerformance()

The generated performance.

Returns:

the performance, can be null

getBestClassifier

public weka.classifiers.Classifier getBestClassifier()

Returns the best classifier found. For optimizers like GridSearch, this outputs the best setup found. For all others, that's the same setup as the actual classifier that was evaluated.

Returns:

the best setup or the original setup

getOutputBestSetup

public boolean getOutputBestSetup()

Returns whether the best setup is output in case of optimizers like GridSearch/MultiSearch.

Returns:

true if the best setup is output

preProcessCheck

protected String preProcessCheck()

Checks whether all pre-conditions have been met.

Specified by:

preProcessCheck in class AbstractJob

Returns:

null if everything is OK, otherwise an error message

getBestClassifier

protected weka.classifiers.Classifier getBestClassifier(weka.classifiers.Classifier template,
                                                        weka.classifiers.Classifier trained)

In case of GridSearch/MultiSearch the best setup is returned, otherwise the classifier itself.

Parameters:

template - the template classifier

trained - the trained classifier

Returns:

either the best classifier (in case of GridSearch/MultiSearch) or the template

process

protected void process()
                throws Exception

Does the actual execution of the job.

Specified by:

process in class AbstractJob

Throws:

Exception - if fails to execute job

stopExecution

public void stopExecution()

Stops the execution.

Specified by:

stopExecution in interface Stoppable

Overrides:

stopExecution in class AbstractJob

postProcessCheck

protected String postProcessCheck()

Checks whether all post-conditions have been met.

Specified by:

postProcessCheck in class AbstractJob

Returns:

null if everything is OK, otherwise an error message

cleanUp

public void cleanUp()

Cleans up data structures, frees up memory. Sets the input data to null.

Specified by:

cleanUp in interface CleanUpHandler

Specified by:

cleanUp in interface Job

Overrides:

cleanUp in class AbstractJob

getAdditionalErrorInformation

protected String getAdditionalErrorInformation()

Returns additional information to be added to the error message.

Overrides:

getAdditionalErrorInformation in class AbstractJob

Returns:

the additional information

toString

public String toString()

Returns a string representation of the job.

Specified by:

toString in class AbstractJob

Returns:

a string representation