org.jpmml.rexp.HurdleConverter Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of pmml-rexp Show documentation
Show all versions of pmml-rexp Show documentation
JPMML R to PMML converter
The newest version!
/*
* Copyright (c) 2024 Villu Ruusmann
*
* This file is part of JPMML-R
*
* JPMML-R is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* JPMML-R is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with JPMML-R. If not, see {
private ContinuousLabel label = null;
public HurdleConverter(RGenericVector hurdle){
super(hurdle);
}
@Override
public PMML encodePMML(RExpEncoder encoder){
Model zeroModel = encodeModel(HurdleConverter.NAME_ZERO, encoder);
OutputField zeroPredictField = ModelUtil.createPredictedField(FieldNameUtil.create("predict", "zero"), OpType.CONTINUOUS, DataType.DOUBLE);
encoder.createDerivedField(zeroModel, zeroPredictField, true);
Segment zeroSegment = new Segment(True.INSTANCE, zeroModel)
.setId("zero");
Model countModel = encodeModel(HurdleConverter.NAME_COUNT, encoder);
OutputField countPredictField = ModelUtil.createPredictedField(FieldNameUtil.create("predict", "count"), OpType.CONTINUOUS, DataType.DOUBLE);
encoder.createDerivedField(countModel, countPredictField, true);
Segment countSegment = new Segment(True.INSTANCE, countModel)
.setId("count");
Expression adjExpression = ExpressionUtil.createApply(PMMLFunctions.EXP,
ExpressionUtil.createApply(PMMLFunctions.SUBTRACT,
ExpressionUtil.createApply(PMMLFunctions.LN, new FieldRef(zeroPredictField)),
ExpressionUtil.createApply(RExpFunctions.STATS_PPOIS, ExpressionUtil.createConstant(0), new FieldRef(countPredictField))
)
);
DerivedField adjZeroPredictField = encoder.createDerivedField(FieldNameUtil.create("adjusted", zeroPredictField.requireName()), OpType.CONTINUOUS, DataType.DOUBLE, adjExpression);
Expression targetExpression = ExpressionUtil.createApply(PMMLFunctions.MULTIPLY, new FieldRef(countPredictField), new FieldRef(adjZeroPredictField));
DerivedField targetField = encoder.createDerivedField(FieldNameUtil.create("adjusted", countPredictField.requireName()), OpType.CONTINUOUS, DataType.DOUBLE, targetExpression);
Feature feature = new ContinuousFeature(encoder, targetField);
Schema schema = new Schema(encoder, this.label, Collections.emptyList());
RegressionModel fullModel = RegressionModelUtil.createRegression(Collections.singletonList(feature), Collections.singletonList(1d), null, RegressionModel.NormalizationMethod.NONE, schema);
OutputField truncatedTargetField = new OutputField(FieldNameUtil.create("truncated", this.label.getName()), OpType.CONTINUOUS, DataType.DOUBLE)
.setResultFeature(ResultFeature.TRANSFORMED_VALUE)
.setExpression(new FieldRef(countPredictField));
Output output = new Output()
.addOutputFields(truncatedTargetField);
fullModel.setOutput(output);
Segment fullSegment = new Segment(True.INSTANCE, fullModel)
.setId("full");
List models = Arrays.asList(zeroModel, countModel, fullModel);
Segmentation segmentation = new Segmentation(Segmentation.MultipleModelMethod.MODEL_CHAIN, null)
.addSegments(zeroSegment, countSegment, fullSegment);
MiningModel miningModel = new MiningModel(MiningFunction.REGRESSION, MiningModelUtil.createMiningSchema(models))
.setSegmentation(segmentation);
return encoder.encodePMML(miningModel);
}
private Model encodeModel(String name, RExpEncoder encoder){
RGenericVector hurdle = getObject();
RDoubleVector coefficients = hurdle.getGenericElement("coefficients").getDoubleElement(name);
RStringVector dist = hurdle.getGenericElement("dist").getStringElement(name);
RExp terms = hurdle.getGenericElement("terms").getElement(name);
RGenericVector model = hurdle.getGenericElement("model");
RStringVector coefficientNames = coefficients.names();
FormulaContext context = new ModelFrameFormulaContext(model);
Formula formula = FormulaUtil.createFormula(terms, context, encoder);
switch(name){
case HurdleConverter.NAME_COUNT:
FormulaUtil.setLabel(formula, terms, null, encoder);
ContinuousLabel continuousLabel = (ContinuousLabel)encoder.getLabel();
// XXX
DataField dataField = (DataField)encoder.getField(continuousLabel.getName());
dataField.setDataType(DataType.DOUBLE);
this.label = new ContinuousLabel(dataField);
break;
case HurdleConverter.NAME_ZERO:
break;
default:
throw new IllegalArgumentException();
}
encoder.setLabel(new ContinuousLabel(DataType.DOUBLE));
List features = encoder.getFeatures();
if(!features.isEmpty()){
features.clear();
}
List names = FormulaUtil.removeSpecialSymbol(coefficientNames.getDequotedValues(), "(Intercept)");
FormulaUtil.addFeatures(formula, names, true, encoder);
features = encoder.getFeatures();
Schema schema = encoder.createSchema();
Double intercept = coefficients.getElement("(Intercept)", false);
SchemaUtil.checkSize(coefficients.size() - (intercept != null ? 1 : 0), features);
List featureCoefficients = new ArrayList<>();
for(Feature feature : features){
Double coefficient = formula.getCoefficient(feature, coefficients);
featureCoefficients.add(coefficient);
}
RegressionModel regressionModel;
switch(name){
case HurdleConverter.NAME_ZERO:
switch(dist.asScalar()){
case "binomial":
regressionModel = RegressionModelUtil.createRegression(features, featureCoefficients, intercept, RegressionModel.NormalizationMethod.LOGIT, schema);
break;
default:
throw new IllegalArgumentException();
}
break;
case HurdleConverter.NAME_COUNT:
switch(dist.asScalar()){
case "poisson":
regressionModel = RegressionModelUtil.createRegression(features, featureCoefficients, intercept, RegressionModel.NormalizationMethod.EXP, schema);
break;
default:
throw new IllegalArgumentException();
}
break;
default:
throw new IllegalArgumentException(name);
}
return regressionModel;
}
private static final String NAME_COUNT = "count";
private static final String NAME_ZERO = "zero";
}