weka.knowledgeflow.steps

Class PredictionAppender

java.lang.Object

weka.knowledgeflow.steps.BaseStep

weka.knowledgeflow.steps.PredictionAppender

All Implemented Interfaces:

java.io.Serializable, BaseStepExtender, Step

@KFStep(name="PredictionAppender",
        category="Evaluation",
        toolTipText="Append predictions from classifiers or clusterers to incoming data ",
        iconPath="weka/gui/knowledgeflow/icons/PredictionAppender.gif")
public class PredictionAppender
extends BaseStep

Step that can produce data with predictions appended from batch or incremental classifiers and clusterers

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	m_appendProbabilities True if probabilities are to be appended
protected Data	m_instanceData Re-usable Data object for streaming output
protected Instances	m_streamingOutputStructure Holds structure of streaming output
protected java.util.List<java.lang.Integer>	m_stringAttIndexes Keep track of indexes of string attributes in the streaming case

Fields inherited from class weka.knowledgeflow.steps.BaseStep

m_stepIsResourceIntensive, m_stepManager, m_stepName

Constructor Summary

Constructors

Constructor and Description
PredictionAppender()

Method Summary

Modifier and Type	Method and Description
boolean	getAppendProbabilities() Get whether to append probability distributions rather than predicted classes
java.util.List<java.lang.String>	getIncomingConnectionTypes() Get the incoming connection types that this step accepts
java.util.List<java.lang.String>	getOutgoingConnectionTypes() Get a list of outgoing connection types that this step can produce at this time
protected Instances	makeOutputDataClassifier(Instances inputData, Classifier classifier, boolean distribution, java.lang.String relationNameModifier) Make an output dataset for a classifier.
protected Instances	makeOutputDataClusterer(Instances inputData, Clusterer clusterer, boolean distribution, java.lang.String relationNameModifier) Make an output dataset for a clusterer.
protected void	predictLabelClassifier(Classifier classifier, Instance inst, Instance instOrig) Add a label to an instance using a classifier
protected void	predictLabelClusterer(Clusterer clusterer, Instance inst, Instance instOrig) Add a cluster label to an instance using a clusterer
protected void	predictProbabilitiesClassifier(Classifier classifier, Instance inst, Instance instOrig) Add a distribution over class labels to an instance using a classifier
protected void	predictProbabilitiesClusterer(DensityBasedClusterer clusterer, Instance inst, Instance instOrig) Add a distribution over cluster labels to an instance using a DensityBasedClusterer
protected void	processBatchClassifierCase(Data data, Instances trainingData, Instances testData) Process a batch classifier
protected void	processBatchClustererCase(Data data, Instances trainingData, Instances testData) Process a batch classifier
void	processIncoming(Data data) Process incoming data
protected void	processIncrementalClassifier(Data data, Instance inst) Process an incremental classifier
void	setAppendProbabilities(boolean append) Set whether to append probability distributions rather than predicted classes
void	stepInit() Initialize the step

Methods inherited from class weka.knowledgeflow.steps.BaseStep

environmentSubstitute, getCustomEditorForStep, getDefaultSettings, getInteractiveViewers, getInteractiveViewersImpls, getName, getStepManager, globalInfo, isResourceIntensive, isStopRequested, outputStructureForConnectionType, setName, setStepIsResourceIntensive, setStepManager, setStepMustRunSingleThreaded, start, stepMustRunSingleThreaded, stop

Methods inherited from class java.lang.Object

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

Field Detail

m_appendProbabilities

protected boolean m_appendProbabilities

True if probabilities are to be appended

m_streamingOutputStructure

protected Instances m_streamingOutputStructure

Holds structure of streaming output

m_instanceData

protected Data m_instanceData

Re-usable Data object for streaming output

m_stringAttIndexes

protected java.util.List<java.lang.Integer> m_stringAttIndexes

Keep track of indexes of string attributes in the streaming case

Constructor Detail

PredictionAppender

public PredictionAppender()

Method Detail

stepInit

public void stepInit()
              throws WekaException

Initialize the step

Throws:

WekaException - if a problem occurs

getIncomingConnectionTypes

public java.util.List<java.lang.String> getIncomingConnectionTypes()

Get the incoming connection types that this step accepts

Returns:

a list of acceptable incoming connection types

getOutgoingConnectionTypes

public java.util.List<java.lang.String> getOutgoingConnectionTypes()

Get a list of outgoing connection types that this step can produce at this time

Returns:

a list of outgoing connection types

processIncoming

public void processIncoming(Data data)
                     throws WekaException

Process incoming data

Specified by:

processIncoming in interface BaseStepExtender

Specified by:

processIncoming in interface Step

Overrides:

processIncoming in class BaseStep

Parameters:

data - the Data object to process

Throws:

WekaException - if a problem occurs

processIncrementalClassifier

protected void processIncrementalClassifier(Data data,
                                            Instance inst)
                                     throws WekaException

Process an incremental classifier

Parameters:

data - the Data object to process

inst - the instance to process

Throws:

WekaException - if a problem occurs

processBatchClustererCase

protected void processBatchClustererCase(Data data,
                                         Instances trainingData,
                                         Instances testData)
                                  throws WekaException

Process a batch classifier

Parameters:

data - the Data object to process

trainingData - the training data (can be null)

testData - the test data (can be null)

Throws:

WekaException - if a problem occurs

processBatchClassifierCase

protected void processBatchClassifierCase(Data data,
                                          Instances trainingData,
                                          Instances testData)
                                   throws WekaException

Process a batch classifier

Parameters:

data - the Data object to process

trainingData - the training data (can be null)

testData - the test data (can be null)

Throws:

WekaException - if a problem occurs

predictLabelClusterer

protected void predictLabelClusterer(Clusterer clusterer,
                                     Instance inst,
                                     Instance instOrig)
                              throws WekaException

Add a cluster label to an instance using a clusterer

Parameters:

clusterer - the clusterer to use

inst - the instance to append a prediction to

instOrig - the original instance

Throws:

WekaException - if a problem occurs

predictProbabilitiesClusterer

protected void predictProbabilitiesClusterer(DensityBasedClusterer clusterer,
                                             Instance inst,
                                             Instance instOrig)
                                      throws WekaException

Add a distribution over cluster labels to an instance using a DensityBasedClusterer

Parameters:

clusterer - the clusterer to use

inst - the instance to append a prediction to

instOrig - the original instance

Throws:

WekaException - if a problem occurs

predictLabelClassifier

protected void predictLabelClassifier(Classifier classifier,
                                      Instance inst,
                                      Instance instOrig)
                               throws WekaException

Add a label to an instance using a classifier

Parameters:

classifier - the classifier to use

inst - the instance to append prediction to

instOrig - the original instance

Throws:

WekaException - if a problem occurs

predictProbabilitiesClassifier

protected void predictProbabilitiesClassifier(Classifier classifier,
                                              Instance inst,
                                              Instance instOrig)
                                       throws WekaException

Add a distribution over class labels to an instance using a classifier

Parameters:

classifier - the classifier to use

inst - the instance to append prediction to

instOrig - the original instance

Throws:

WekaException - if a problem occurs

makeOutputDataClusterer

protected Instances makeOutputDataClusterer(Instances inputData,
                                            Clusterer clusterer,
                                            boolean distribution,
                                            java.lang.String relationNameModifier)
                                     throws java.lang.Exception

Make an output dataset for a clusterer. Either a single attribute is added for holding cluster labels, or a series of attributes are added in order to hold predicted cluster distribution

Parameters:

inputData - the incoming data

clusterer - the clusterer

distribution - true if a distribution over cluster labels will be predicted

relationNameModifier - modifier to add to the incoming relation name

Returns:

the output dataset

Throws:

java.lang.Exception - if a problem occurs

makeOutputDataClassifier

protected Instances makeOutputDataClassifier(Instances inputData,
                                             Classifier classifier,
                                             boolean distribution,
                                             java.lang.String relationNameModifier)
                                      throws java.lang.Exception

Make an output dataset for a classifier. Either a single attribute is added for holding class labels, or a series of attributes are added in order to hold predicted class distribution

Parameters:

inputData - the incoming data

classifier - the classifier

distribution - true if a distribution over class labels will be predicted

relationNameModifier - modifier to add to the incoming relation name

Returns:

the output dataset

Throws:

java.lang.Exception - if a problem occurs

setAppendProbabilities

public void setAppendProbabilities(boolean append)

Set whether to append probability distributions rather than predicted classes

Parameters:

append - true to append probability distributions

getAppendProbabilities

@OptionMetadata(displayName="Append probabilities",
                description="Append probabilities")
public boolean getAppendProbabilities()

Get whether to append probability distributions rather than predicted classes

Returns:

true if probability distributions are to be appended