com.rapidminer.operator.learner.meta
Class AbstractMetaLearner

java.lang.Object
  com.rapidminer.tools.AbstractObservable<Operator>
      com.rapidminer.operator.Operator
          com.rapidminer.operator.OperatorChain
              com.rapidminer.operator.learner.meta.AbstractMetaLearner

All Implemented Interfaces:

ConfigurationListener, PreviewListener, ResourceConsumer, CapabilityProvider, Learner, ParameterHandler, LoggingHandler, Observable<Operator>

Direct Known Subclasses:

AdaBoost, AdditiveRegression, Bagging, BayBoostStream, BayesianBoosting, Binary2MultiClassLearner, ClassificationByRegression, CostBasedThresholdLearner, HierarchicalLearner, HierarchicalMultiClassLearner, MetaCost, MultiwayDecisionTree, RelativeRegression, TransformedRegression, Tree2RuleConverter

public abstract class AbstractMetaLearner
extends OperatorChain
implements Learner

A MetaLearner is an operator that encapsulates one or more learning steps to build its model. New meta learning schemes should extend this class to support the same parameters as other learners. The main purpose of this class is to perform some compatibility checks.

Author:

Ingo Mierswa

Field Summary

protected InputPort	exampleSetInput
protected InputPort	innerModelSink

Fields inherited from interface com.rapidminer.operator.learner.CapabilityProvider

PROPERTY_RAPIDMINER_GENERAL_CAPABILITIES_WARN

Constructor Summary

AbstractMetaLearner(OperatorDescription description)

Method Summary

protected Model	applyInnerLearner(ExampleSet exampleSet) This is a convenience method to apply the inner operators and return the model which must be output of the last operator.
void	doWork() Trains a model using an ExampleSet from the input.
protected void	executeInnerLearner()
PerformanceVector	getEstimatedPerformance() The default implementation throws an exception.
InputPort	getInnerModelSink()
OutputPort	getModelOutputPort()
InputPort	getTrainingSetInputPort()
AttributeWeights	getWeights(ExampleSet exampleSet) Returns the calculated weight vectors.
protected MetaData	modifyExampleSetMetaData(ExampleSetMetaData unmodifiedMetaData) This method can be used by subclasses to additionally change the example set meta data delivered to the inner learner
protected MetaData	modifyGeneratedModelMetaData(PredictionModelMetaData unmodifiedMetaData) Modifies the meta data of the generated model.
boolean	shouldAutoConnect(OutputPort port) By default, all ports will be auto-connected by ExecutionUnit.autoWire(CompatibilityLevel, boolean, boolean).
boolean	shouldCalculateWeights() Returns true if the user wants to calculate feature weights (depending on a parameter).
boolean	shouldEstimatePerformance() Returns true if the user wants to estimate the performance (depending on a parameter).

Methods inherited from class com.rapidminer.operator.OperatorChain

addOperator, addOperator, addSubprocess, areSubprocessesExtendable, assumePreconditionsSatisfied, checkDeprecations, checkIO, checkNumberOfInnerOperators, checkProperties, clear, cloneOperator, collectErrors, createProcessTree, createSubprocess, freeMemory, getAllInnerOperators, getAllInnerOperatorsAndMe, getImmediateChildren, getIndexOfOperator, getInnerOperatorCondition, getMaxNumberOfInnerOperators, getMinNumberOfInnerOperators, getNumberOfAllOperators, getNumberOfOperators, getNumberOfSubprocesses, getOperator, getOperatorFromAll, getOperators, getSubprocess, getSubprocesses, isEnabled, lookupOperator, notifyRenaming, performAdditionalChecks, processFinished, processStarts, propagateDirtyness, registerOperator, removeOperator, removeSubprocess, shouldAddNonConsumedInput, shouldReturnInnerOutput, unregisterOperator, updateExecutionOrder, walk

Methods inherited from class com.rapidminer.operator.Operator

acceptsInput, addError, addError, addValue, addWarning, apply, apply, checkAll, checkAllExcludingMetaData, checkForStop, clearErrorList, createExperimentTree, createExperimentTree, createFromXML, createFromXML, createFromXML, createMarkedExperimentTree, createMarkedProcessTree, createProcessTree, disconnectPorts, execute, fireUpdate, getAddOnlyAdditionalOutput, getApplyCount, getCompatibilityLevel, getDeliveredOutputClasses, getDeprecationInfo, getDesiredInputClasses, getDOMRepresentation, getEncoding, getErrorList, getExecutionUnit, getExperiment, getIncompatibleVersionChanges, getInput, getInput, getInput, getInputClasses, getInputDescription, getInputPorts, getIODescription, getLog, getLogger, getName, getNumberOfBreakpoints, getOperatorClassName, getOperatorDescription, getOutputClasses, getOutputPorts, getParameter, getParameterAsBoolean, getParameterAsChar, getParameterAsColor, getParameterAsDouble, getParameterAsFile, getParameterAsFile, getParameterAsInputStream, getParameterAsInt, getParameterAsMatrix, getParameterAsRepositoryLocation, getParameterAsString, getParameterHandler, getParameterList, getParameters, getParameterTupel, getParameterType, getParameterTypes, getParent, getPortOwner, getProcess, getResourceConsumptionEstimator, getRoot, getStartTime, getTransformer, getUserDescription, getValue, getValues, getXML, getXML, getXML, hasBreakpoint, hasBreakpoint, hasInput, inApplyLoop, isDebugMode, isDirty, isExpanded, isParallel, isParameterSet, isRunning, log, log, logError, logNote, logWarning, makeDirty, makeDirtyOnUpdate, preAutoWire, producesOutput, register, remove, removeAndKeepConnections, rename, resume, setBreakpoint, setCompatibilityLevel, setEnabled, setEnclosingProcess, setExpanded, setInput, setListParameter, setPairParameter, setParameter, setParameters, setUserDescription, shouldAutoConnect, shouldStopStandaloneExecution, toString, transformMetaData, writeXML, writeXML

Methods inherited from class com.rapidminer.tools.AbstractObservable

addObserver, addObserverAsFirst, fireUpdate, removeObserver

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.rapidminer.operator.learner.Learner

getName, learn

Methods inherited from interface com.rapidminer.operator.learner.CapabilityProvider

supportsCapability

Field Detail

exampleSetInput

protected final InputPort exampleSetInput

innerModelSink

protected final InputPort innerModelSink

Constructor Detail

AbstractMetaLearner

public AbstractMetaLearner(OperatorDescription description)

Method Detail

modifyGeneratedModelMetaData

protected MetaData modifyGeneratedModelMetaData(PredictionModelMetaData unmodifiedMetaData)

Modifies the meta data of the generated model.

modifyExampleSetMetaData

protected MetaData modifyExampleSetMetaData(ExampleSetMetaData unmodifiedMetaData)

This method can be used by subclasses to additionally change the example set meta data delivered to the inner learner

getTrainingSetInputPort

public InputPort getTrainingSetInputPort()

getModelOutputPort

public OutputPort getModelOutputPort()

getInnerModelSink

public InputPort getInnerModelSink()

doWork

public void doWork()
            throws OperatorException

Trains a model using an ExampleSet from the input. Uses the method learn(ExampleSet).

Overrides:

doWork in class OperatorChain

Throws:

OperatorException

applyInnerLearner

protected Model applyInnerLearner(ExampleSet exampleSet)
                           throws OperatorException

This is a convenience method to apply the inner operators and return the model which must be output of the last operator.

Throws:

OperatorException

executeInnerLearner

protected void executeInnerLearner()
                            throws OperatorException

Throws:

OperatorException

shouldAutoConnect

public boolean shouldAutoConnect(OutputPort port)

Description copied from class: Operator

By default, all ports will be auto-connected by ExecutionUnit.autoWire(CompatibilityLevel, boolean, boolean). Optional outputs were handled up to version 4.4 by parameters. From 5.0 on, optional outputs are computed iff the corresponding port is connected. For backward compatibility, operators can check if we should auto-connect a port by overriding this method (e.g. by checking a deprecated parameter). TODO: Remove in later versions

Overrides:

shouldAutoConnect in class Operator

shouldEstimatePerformance

public boolean shouldEstimatePerformance()

Returns true if the user wants to estimate the performance (depending on a parameter). In this case the method getEstimatedPerformance() must also be overridden and deliver the estimated performance. The default implementation returns false.

Specified by:

shouldEstimatePerformance in interface Learner

shouldCalculateWeights

public boolean shouldCalculateWeights()

Returns true if the user wants to calculate feature weights (depending on a parameter). In this case the method getWeights() must also be overriden and deliver the calculated weights. The default implementation returns false.

Specified by:

shouldCalculateWeights in interface Learner

getEstimatedPerformance

public PerformanceVector getEstimatedPerformance()
                                          throws OperatorException

The default implementation throws an exception.

Specified by:

getEstimatedPerformance in interface Learner

Throws:

OperatorException

getWeights

public AttributeWeights getWeights(ExampleSet exampleSet)
                            throws OperatorException

Returns the calculated weight vectors. The default implementation throws an exception.

Specified by:

getWeights in interface Learner

Throws:

OperatorException