Package org.jbpm.prediction.pmml

Class AbstractPMMLBackend

java.lang.Object

org.jbpm.prediction.pmml.AbstractPMMLBackend

All Implemented Interfaces:

org.kie.internal.task.api.prediction.PredictionService

Direct Known Subclasses:

PMMLRandomForest

public abstract class AbstractPMMLBackend
extends Object
implements org.kie.internal.task.api.prediction.PredictionService

Field Summary

Fields

Modifier and Type	Field	Description
protected double	confidenceThreshold
protected List<String>	inputFeatures
protected String	outcomeFeatureName
protected List<? extends org.jpmml.evaluator.OutputField>	outputFields

Constructor Summary

Constructors

Constructor	Description
AbstractPMMLBackend(List<String> inputFeatures, String outputFeatureName, double confidenceThreshold, File pmmlFile)

Method Summary

Modifier and Type	Method	Description
protected Map<String,?>	evaluate(Map<String,Object> data)
protected abstract Map<String,Object>	preProcess(Map<String,Object> data)	Returns the processed data (e.g. perform categorisation, etc).
void	train(org.kie.api.task.model.Task task, Map<String,Object> inputData, Map<String,Object> outputData)	Method to train a model.

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.kie.internal.task.api.prediction.PredictionService

getIdentifier, predict

Field Detail

outputFields

protected final List<? extends org.jpmml.evaluator.OutputField> outputFields

inputFeatures

protected List<String> inputFeatures

outcomeFeatureName

protected String outcomeFeatureName

confidenceThreshold

protected double confidenceThreshold

Constructor Detail

AbstractPMMLBackend

public AbstractPMMLBackend(List<String> inputFeatures,
                           String outputFeatureName,
                           double confidenceThreshold,
                           File pmmlFile)

Method Detail

train

public void train(org.kie.api.task.model.Task task,
                  Map<String,Object> inputData,
                  Map<String,Object> outputData)

Method to train a model. In the PMML case, this is a no-op.

Specified by:

train in interface org.kie.internal.task.api.prediction.PredictionService

Parameters:

task - Human task data

inputData - A map containing the input attribute names as keys and the attribute values as values.

outputData - A map containing the output attribute names as keys and the attribute values as values.

preProcess

protected abstract Map<String,Object> preProcess(Map<String,Object> data)

Returns the processed data (e.g. perform categorisation, etc). If no processing is needed, simply return the original data.

Parameters:

data - A map containing the input data, with attribute names as key and values as values.

Returns:

data A map containing the processed data, with attribute names as key and values as values.

evaluate

protected Map<String,?> evaluate(Map<String,Object> data)