Package adams.opt.genetic

Class AbstractClassifierBasedGeneticAlgorithm.ClassifierBasedGeneticAlgorithmJob<T extends AbstractClassifierBasedGeneticAlgorithm>

java.lang.Object

adams.core.logging.LoggingObject

adams.multiprocess.AbstractJob

adams.opt.genetic.AbstractGeneticAlgorithm.GeneticAlgorithmJob<T>

adams.opt.genetic.AbstractClassifierBasedGeneticAlgorithm.ClassifierBasedGeneticAlgorithmJob<T>

All Implemented Interfaces:

CleanUpHandler, LoggingSupporter, SizeOfHandler, Stoppable, StoppableWithFeedback, Job, JobWithOwner<T>, Serializable

Direct Known Subclasses:

AbstractClassifierBasedGeneticAlgorithmWithSecondEvaluation.ClassifierBasedGeneticAlgorithmWithSecondEvaluationJob

Enclosing class:

AbstractClassifierBasedGeneticAlgorithm

public abstract static class AbstractClassifierBasedGeneticAlgorithm.ClassifierBasedGeneticAlgorithmJob<T extends AbstractClassifierBasedGeneticAlgorithm>
extends AbstractGeneticAlgorithm.GeneticAlgorithmJob<T>

Job class for algorithms with datasets.

Author:

dale

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected int	m_ClassLabelIndex	the class label index.
protected weka.core.Instances	m_Data	the data to use.
protected int	m_Folds	the cross-validation folds.
protected Measure	m_Measure	the measure to use for evaluating the fitness.
protected int	m_Seed	the cross-validation seed.
protected weka.core.Instances	m_TestData	the test data to use (can be null).

Fields inherited from class adams.opt.genetic.AbstractGeneticAlgorithm.GeneticAlgorithmJob

m_Chromosome, m_Fitness, m_Genetic, m_Weights

Fields inherited from class adams.multiprocess.AbstractJob

m_Complete, m_ExecutionError, m_JobCompleteListener, m_JobInfo, m_Stopped

Fields inherited from class adams.core.logging.LoggingObject

m_Logger, m_LoggingIsEnabled, m_LoggingLevel

Constructor Summary

Constructors

Constructor	Description
ClassifierBasedGeneticAlgorithmJob(T g, int chromosome, int[] w, weka.core.Instances data, weka.core.Instances testData)	Initializes the job.

Method Summary

Modifier and Type	Method	Description
protected Map<String,Object>	assembleSetup(double fitness, weka.classifiers.Classifier cls, int chromosome, int[] weights)	Assembles the data for the textual setup output.
protected File	createFileName(double fitness, weka.core.Instances data, String ext)	Generates a file name for the fitness.
protected double	evaluateClassifier(weka.classifiers.Classifier cls, weka.core.Instances data, int folds, int seed)	Evaluates the classifier on the dataset and returns the metric.
protected double	evaluateClassifier(weka.classifiers.Classifier cls, weka.core.Instances data, weka.core.Instances test)	Evaluates the classifier on the dataset and returns the metric.
protected void	generateOutput(double fitness, weka.core.Instances data, weka.classifiers.Classifier cls, int chromosome, int[] weights)	Generates the output requested output.
int	getFolds()	Returns the number of cross-validation folds.
protected weka.core.Instances	getInstances()	Returns the instances in use by the genetic algorithm.
Measure	getMeasure()	Returns the measure used for evaluating the fitness.
int	getSeed()	Returns the cross-validation seed.
protected weka.core.Instances	getTestInstances()	Returns the test instances in use by the genetic algorithm.
protected void	outputDataset(double fitness, weka.core.Instances data)	Saves the instances to a file.
protected void	outputSetup(double fitness, weka.core.Instances data, weka.classifiers.Classifier cls, int chromosome, int[] weights)	Saves the setup to a props file.
protected weka.classifiers.Evaluation	postProcess(weka.classifiers.Evaluation eval)	Post-processes the Evaluation if necessary.
protected String	preProcessCheck()	Checks whether all pre-conditions have been met.

Methods inherited from class adams.opt.genetic.AbstractGeneticAlgorithm.GeneticAlgorithmJob

calcNewFitness, cleanUp, getChromosome, getFitness, getOwner, getWeights, postProcessCheck, process, toString, weightsToString, weightsToString

Methods inherited from class adams.multiprocess.AbstractJob

execute, getAdditionalErrorInformation, getExecutionError, getJobCompleteListener, getJobInfo, hasExecutionError, isComplete, isStopped, jobCompleted, setJobCompleteListener, setJobInfo, stopExecution

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.multiprocess.Job

execute, getExecutionError, getJobCompleteListener, getJobInfo, hasExecutionError, isComplete, jobCompleted, setJobCompleteListener

Methods inherited from interface adams.core.logging.LoggingSupporter

getLogger, isLoggingEnabled

Methods inherited from interface adams.core.Stoppable

stopExecution

Methods inherited from interface adams.core.StoppableWithFeedback

isStopped

Field Detail

m_Measure

protected Measure m_Measure

the measure to use for evaluating the fitness.

m_ClassLabelIndex

protected int m_ClassLabelIndex

the class label index.

m_Data

protected weka.core.Instances m_Data

the data to use.

m_TestData

protected weka.core.Instances m_TestData

the test data to use (can be null).

m_Seed

protected int m_Seed

the cross-validation seed.

m_Folds

protected int m_Folds

the cross-validation folds.

Constructor Detail

ClassifierBasedGeneticAlgorithmJob

public ClassifierBasedGeneticAlgorithmJob(T g,
                                          int chromosome,
                                          int[] w,
                                          weka.core.Instances data,
                                          weka.core.Instances testData)

Initializes the job.

Parameters:

g - the algorithm object this job belongs to

chromosome - the chromsome index

w - the initial weights

data - the data to use

testData - the test data to use, null for cross-validation

Method Detail

getInstances

protected weka.core.Instances getInstances()

Returns the instances in use by the genetic algorithm.

Returns:

the instances

getTestInstances

protected weka.core.Instances getTestInstances()

Returns the test instances in use by the genetic algorithm.

Returns:

the instances

getMeasure

public Measure getMeasure()

Returns the measure used for evaluating the fitness.

Returns:

the measure

getSeed

public int getSeed()

Returns the cross-validation seed.

Returns:

the seed

getFolds

public int getFolds()

Returns the number of cross-validation folds.

Returns:

the number of folds

preProcessCheck

protected String preProcessCheck()

Checks whether all pre-conditions have been met.

Overrides:

preProcessCheck in class AbstractGeneticAlgorithm.GeneticAlgorithmJob<T extends AbstractClassifierBasedGeneticAlgorithm>

Returns:

null if everything is OK, otherwise an error message

postProcess

protected weka.classifiers.Evaluation postProcess(weka.classifiers.Evaluation eval)

Post-processes the Evaluation if necessary.

Parameters:

eval - the evaluation to post-process

Returns:

the (potentially) updated evaluation

evaluateClassifier

protected double evaluateClassifier(weka.classifiers.Classifier cls,
                                    weka.core.Instances data,
                                    int folds,
                                    int seed)
                             throws Exception

Evaluates the classifier on the dataset and returns the metric.

Parameters:

cls - the classifier to evaluate

data - the data to use for evaluation

folds - the number of folds to use

seed - the seed for the randomization

Returns:

the metric

Throws:

Exception - if the evaluation fails

evaluateClassifier

protected double evaluateClassifier(weka.classifiers.Classifier cls,
                                    weka.core.Instances data,
                                    weka.core.Instances test)
                             throws Exception

Evaluates the classifier on the dataset and returns the metric.

Parameters:

cls - the classifier to evaluate

data - the data to use for evaluation

test - the test data to use

Returns:

the metric

Throws:

Exception - if the evaluation fails

createFileName

protected File createFileName(double fitness,
                              weka.core.Instances data,
                              String ext)

Generates a file name for the fitness.

Parameters:

fitness - the current fitness

data - the dataset

ext - the extension (not dot!)

Returns:

the file

outputDataset

protected void outputDataset(double fitness,
                             weka.core.Instances data)
                      throws Exception

Saves the instances to a file.

Parameters:

fitness - the current measure/fitness

data - the instances to save

Throws:

Exception - if saving the file fails

assembleSetup

protected Map<String,Object> assembleSetup(double fitness,
                                                 weka.classifiers.Classifier cls,
                                                 int chromosome,
                                                 int[] weights)

Assembles the data for the textual setup output.

Parameters:

fitness - the current fitness

cls - the current classifier

chromosome - the chromosome responsible

weights - the weights

Returns:

the data

outputSetup

protected void outputSetup(double fitness,
                           weka.core.Instances data,
                           weka.classifiers.Classifier cls,
                           int chromosome,
                           int[] weights)
                    throws Exception

Saves the setup to a props file.

Parameters:

fitness - the current measure/fitness

data - the dataset

cls - the current classifier setup

chromosome - the chromosome responsible

weights - the current weights/bits

Throws:

Exception - if saving the file fails

generateOutput

protected void generateOutput(double fitness,
                              weka.core.Instances data,
                              weka.classifiers.Classifier cls,
                              int chromosome,
                              int[] weights)
                       throws Exception

Generates the output requested output.

Parameters:

fitness - the current fitness

data - the dataset

cls - the current classifier

chromosome - the chromosome responsible

weights - the current weights/bits

Throws:

Exception - if output fails