org.apache.storm.pmml.runner.jpmml

Class JPmmlModelRunner

java.lang.Object

org.apache.storm.pmml.runner.jpmml.JPmmlModelRunner

All Implemented Interfaces:

ModelRunner, PmmlModelRunner<Tuple,Map<org.dmg.pmml.FieldName,Object>,Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>,Map<org.dmg.pmml.FieldName,?>>

public class JPmmlModelRunner
extends Object
implements PmmlModelRunner<Tuple,Map<org.dmg.pmml.FieldName,Object>,Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>,Map<org.dmg.pmml.FieldName,?>>

JPMML implementation of PmmlModelRunner. It extracts the raw inputs from the tuple for all ‘active fields’, and builds a tuple with the predicted scores for the ‘predicted fields’ and ‘output fields’. In this implementation all the declared streams will have the same scored tuple.

The ‘predicted’, ‘active’, and ‘output’ fields are extracted from the PMML model.

Constructor Summary

Constructors

Constructor and Description
JPmmlModelRunner(org.jpmml.evaluator.Evaluator evaluator, ModelOutputs modelOutputs)

Method Summary

Modifier and Type	Method and Description
Map<org.dmg.pmml.FieldName,Object>	extractRawInputs(Tuple tuple) Extracts from the tuple the raw inputs that are to be scored according to the predictive model.
List<org.dmg.pmml.FieldName>	getActiveFields()
org.jpmml.evaluator.Evaluator	getEval()
ModelOutputs	getModelOutputs()
List<org.dmg.pmml.FieldName>	getOutputFields()
List<org.dmg.pmml.FieldName>	getPredictedFields()
Map<org.dmg.pmml.FieldName,?>	predictScores(Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue> preProcInputs) Compute the predicted scores from the pre-processed inputs in the step above
Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>	preProcessInputs(Map<org.dmg.pmml.FieldName,Object> rawInputs) Pre process inputs, i.e., remove missing fields, outliers, etc
Map<String,List<Object>>	scoredTuplePerStream(Tuple input) Creates and returns a map with the predicted scores that are to be emitted on each stream.

Methods inherited from class java.lang.Object

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

Constructor Detail

JPmmlModelRunner

public JPmmlModelRunner(org.jpmml.evaluator.Evaluator evaluator,
                        ModelOutputs modelOutputs)

Method Detail

extractRawInputs

public Map<org.dmg.pmml.FieldName,Object> extractRawInputs(Tuple tuple)

Description copied from interface: PmmlModelRunner

Extracts from the tuple the raw inputs that are to be scored according to the predictive model.

Specified by:

extractRawInputs in interface PmmlModelRunner<Tuple,Map<org.dmg.pmml.FieldName,Object>,Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>,Map<org.dmg.pmml.FieldName,?>>

Parameters:

tuple - source from which to extract raw inputs

Returns:

The raw inputs extracted from the tuple for all ‘active fields’

preProcessInputs

public Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue> preProcessInputs(Map<org.dmg.pmml.FieldName,Object> rawInputs)

Description copied from interface: PmmlModelRunner

Pre process inputs, i.e., remove missing fields, outliers, etc

Specified by:

preProcessInputs in interface PmmlModelRunner<Tuple,Map<org.dmg.pmml.FieldName,Object>,Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>,Map<org.dmg.pmml.FieldName,?>>

Parameters:

rawInputs - that are to be preprocessed

Returns:

preprocessed output

predictScores

public Map<org.dmg.pmml.FieldName,?> predictScores(Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue> preProcInputs)

Description copied from interface: PmmlModelRunner

Compute the predicted scores from the pre-processed inputs in the step above

Specified by:

predictScores in interface PmmlModelRunner<Tuple,Map<org.dmg.pmml.FieldName,Object>,Map<org.dmg.pmml.FieldName,org.jpmml.evaluator.FieldValue>,Map<org.dmg.pmml.FieldName,?>>

Parameters:

preProcInputs - that are to be preprocessed

Returns:

predicted scores

scoredTuplePerStream

public Map<String,List<Object>> scoredTuplePerStream(Tuple input)

Description copied from interface: ModelRunner

Creates and returns a map with the predicted scores that are to be emitted on each stream. The keys of this map are the stream ids, and the values the predicted scores. It’s up to the implementation to guarantee that the streams ids match the stream ids defined in ModelOutputs. Namely, the set of keys of the Map<String, List<Object>> returned by this method should be a subset of ModelOutputs.streams()

Specified by:

scoredTuplePerStream in interface ModelRunner

Returns:

the predicted scores for the ‘predicted fields’ and ‘output fields’. All the declared streams will have the same scored tuple.

getEval

public org.jpmml.evaluator.Evaluator getEval()

getActiveFields

public List<org.dmg.pmml.FieldName> getActiveFields()

getPredictedFields

public List<org.dmg.pmml.FieldName> getPredictedFields()

getOutputFields

public List<org.dmg.pmml.FieldName> getOutputFields()

getModelOutputs

public ModelOutputs getModelOutputs()