use of org.dmg.pmml.OutputField in project drools by kiegroup.
the class KiePMMLClassificationTableFactoryTest method getClassificationTableBuilders.
@Test
public void getClassificationTableBuilders() {
RegressionTable regressionTableProf = getRegressionTable(3.5, "professional");
RegressionTable regressionTableCler = getRegressionTable(27.4, "clerical");
OutputField outputFieldCat = getOutputField("CAT-1", ResultFeature.PROBABILITY, "CatPred-1");
OutputField outputFieldNum = getOutputField("NUM-1", ResultFeature.PROBABILITY, "NumPred-0");
OutputField outputFieldPrev = getOutputField("PREV", ResultFeature.PREDICTED_VALUE, null);
String targetField = "targetField";
DataField dataField = new DataField();
dataField.setName(FieldName.create(targetField));
dataField.setOpType(OpType.CATEGORICAL);
DataDictionary dataDictionary = new DataDictionary();
dataDictionary.addDataFields(dataField);
RegressionModel regressionModel = new RegressionModel();
regressionModel.setNormalizationMethod(RegressionModel.NormalizationMethod.CAUCHIT);
regressionModel.addRegressionTables(regressionTableProf, regressionTableCler);
regressionModel.setModelName(getGeneratedClassName("RegressionModel"));
Output output = new Output();
output.addOutputFields(outputFieldCat, outputFieldNum, outputFieldPrev);
regressionModel.setOutput(output);
MiningField miningField = new MiningField();
miningField.setUsageType(MiningField.UsageType.TARGET);
miningField.setName(dataField.getName());
MiningSchema miningSchema = new MiningSchema();
miningSchema.addMiningFields(miningField);
regressionModel.setMiningSchema(miningSchema);
PMML pmml = new PMML();
pmml.setDataDictionary(dataDictionary);
pmml.addModels(regressionModel);
final CommonCompilationDTO<RegressionModel> source = CommonCompilationDTO.fromGeneratedPackageNameAndFields(PACKAGE_NAME, pmml, regressionModel, new HasClassLoaderMock());
final RegressionCompilationDTO compilationDTO = RegressionCompilationDTO.fromCompilationDTORegressionTablesAndNormalizationMethod(source, regressionModel.getRegressionTables(), regressionModel.getNormalizationMethod());
Map<String, KiePMMLTableSourceCategory> retrieved = KiePMMLClassificationTableFactory.getClassificationTableBuilders(compilationDTO);
assertNotNull(retrieved);
assertEquals(3, retrieved.size());
retrieved.values().forEach(kiePMMLTableSourceCategory -> commonValidateKiePMMLRegressionTable(kiePMMLTableSourceCategory.getSource()));
Map<String, String> sources = retrieved.entrySet().stream().collect(Collectors.toMap(Map.Entry::getKey, stringKiePMMLTableSourceCategoryEntry -> stringKiePMMLTableSourceCategoryEntry.getValue().getSource()));
commonValidateCompilation(sources);
}
use of org.dmg.pmml.OutputField in project drools by kiegroup.
the class KiePMMLClassificationTableFactoryTest method getClassificationTable.
@Test
public void getClassificationTable() {
RegressionTable regressionTableProf = getRegressionTable(3.5, "professional");
RegressionTable regressionTableCler = getRegressionTable(27.4, "clerical");
OutputField outputFieldCat = getOutputField("CAT-1", ResultFeature.PROBABILITY, "CatPred-1");
OutputField outputFieldNum = getOutputField("NUM-1", ResultFeature.PROBABILITY, "NumPred-0");
OutputField outputFieldPrev = getOutputField("PREV", ResultFeature.PREDICTED_VALUE, null);
String targetField = "targetField";
DataField dataField = new DataField();
dataField.setName(FieldName.create(targetField));
dataField.setOpType(OpType.CATEGORICAL);
DataDictionary dataDictionary = new DataDictionary();
dataDictionary.addDataFields(dataField);
RegressionModel regressionModel = new RegressionModel();
regressionModel.setNormalizationMethod(RegressionModel.NormalizationMethod.CAUCHIT);
regressionModel.addRegressionTables(regressionTableProf, regressionTableCler);
regressionModel.setModelName(getGeneratedClassName("RegressionModel"));
Output output = new Output();
output.addOutputFields(outputFieldCat, outputFieldNum, outputFieldPrev);
regressionModel.setOutput(output);
MiningField targetMiningField = new MiningField();
targetMiningField.setUsageType(MiningField.UsageType.TARGET);
targetMiningField.setName(dataField.getName());
MiningSchema miningSchema = new MiningSchema();
miningSchema.addMiningFields(targetMiningField);
regressionModel.setMiningSchema(miningSchema);
PMML pmml = new PMML();
pmml.setDataDictionary(dataDictionary);
pmml.addModels(regressionModel);
final CommonCompilationDTO<RegressionModel> source = CommonCompilationDTO.fromGeneratedPackageNameAndFields(PACKAGE_NAME, pmml, regressionModel, new HasClassLoaderMock());
final RegressionCompilationDTO compilationDTO = RegressionCompilationDTO.fromCompilationDTORegressionTablesAndNormalizationMethod(source, regressionModel.getRegressionTables(), regressionModel.getNormalizationMethod());
KiePMMLClassificationTable retrieved = KiePMMLClassificationTableFactory.getClassificationTable(compilationDTO);
assertNotNull(retrieved);
assertEquals(regressionModel.getRegressionTables().size(), retrieved.getCategoryTableMap().size());
regressionModel.getRegressionTables().forEach(regressionTable -> assertTrue(retrieved.getCategoryTableMap().containsKey(regressionTable.getTargetCategory().toString())));
assertEquals(regressionModel.getNormalizationMethod().value(), retrieved.getRegressionNormalizationMethod().getName());
assertEquals(OP_TYPE.CATEGORICAL, retrieved.getOpType());
boolean isBinary = regressionModel.getRegressionTables().size() == 2;
assertEquals(isBinary, retrieved.isBinary());
assertEquals(isBinary, retrieved.isBinary());
assertEquals(targetMiningField.getName().getValue(), retrieved.getTargetField());
}
use of org.dmg.pmml.OutputField in project drools by kiegroup.
the class KiePMMLClassificationTableFactoryTest method setStaticGetter.
@Test
public void setStaticGetter() throws IOException {
String variableName = "variableName";
RegressionTable regressionTableProf = getRegressionTable(3.5, "professional");
RegressionTable regressionTableCler = getRegressionTable(27.4, "clerical");
OutputField outputFieldCat = getOutputField("CAT-1", ResultFeature.PROBABILITY, "CatPred-1");
OutputField outputFieldNum = getOutputField("NUM-1", ResultFeature.PROBABILITY, "NumPred-0");
OutputField outputFieldPrev = getOutputField("PREV", ResultFeature.PREDICTED_VALUE, null);
String targetField = "targetField";
DataField dataField = new DataField();
dataField.setName(FieldName.create(targetField));
dataField.setOpType(OpType.CATEGORICAL);
DataDictionary dataDictionary = new DataDictionary();
dataDictionary.addDataFields(dataField);
RegressionModel regressionModel = new RegressionModel();
regressionModel.setNormalizationMethod(RegressionModel.NormalizationMethod.CAUCHIT);
regressionModel.addRegressionTables(regressionTableProf, regressionTableCler);
regressionModel.setModelName(getGeneratedClassName("RegressionModel"));
Output output = new Output();
output.addOutputFields(outputFieldCat, outputFieldNum, outputFieldPrev);
regressionModel.setOutput(output);
MiningField miningField = new MiningField();
miningField.setUsageType(MiningField.UsageType.TARGET);
miningField.setName(dataField.getName());
MiningSchema miningSchema = new MiningSchema();
miningSchema.addMiningFields(miningField);
regressionModel.setMiningSchema(miningSchema);
PMML pmml = new PMML();
pmml.setDataDictionary(dataDictionary);
pmml.addModels(regressionModel);
final CommonCompilationDTO<RegressionModel> source = CommonCompilationDTO.fromGeneratedPackageNameAndFields(PACKAGE_NAME, pmml, regressionModel, new HasClassLoaderMock());
final RegressionCompilationDTO compilationDTO = RegressionCompilationDTO.fromCompilationDTORegressionTablesAndNormalizationMethod(source, regressionModel.getRegressionTables(), regressionModel.getNormalizationMethod());
final LinkedHashMap<String, KiePMMLTableSourceCategory> regressionTablesMap = new LinkedHashMap<>();
regressionModel.getRegressionTables().forEach(regressionTable -> {
String key = "defpack." + regressionTable.getTargetCategory().toString().toUpperCase();
KiePMMLTableSourceCategory value = new KiePMMLTableSourceCategory("", regressionTable.getTargetCategory().toString());
regressionTablesMap.put(key, value);
});
final MethodDeclaration staticGetterMethod = STATIC_GETTER_METHOD.clone();
KiePMMLClassificationTableFactory.setStaticGetter(compilationDTO, regressionTablesMap, staticGetterMethod, variableName);
String text = getFileContent(TEST_02_SOURCE);
MethodDeclaration expected = JavaParserUtils.parseMethod(text);
assertTrue(JavaParserUtils.equalsNode(expected, staticGetterMethod));
}
use of org.dmg.pmml.OutputField in project streamline by hortonworks.
the class MLModelRegistryService method doGetOutputFieldsForPMMLStream.
private List<MLModelField> doGetOutputFieldsForPMMLStream(String pmmlContents) throws SAXException, JAXBException, UnsupportedEncodingException {
List<MLModelField> fieldNames = new ArrayList<>();
PMMLManager pmmlManager = new PMMLManager(IOUtil.unmarshal(new ByteArrayInputStream(pmmlContents.getBytes("UTF-8"))));
Evaluator modelEvaluator = (ModelEvaluator<?>) pmmlManager.getModelManager(null, ModelEvaluatorFactory.getInstance());
modelEvaluator.getPredictedFields().forEach((f) -> fieldNames.add(getModelField(modelEvaluator.getDataField(f))));
modelEvaluator.getOutputFields().forEach((f) -> {
OutputField outputField = modelEvaluator.getOutputField(f);
ResultFeatureType resultFeatureType = outputField.getFeature();
if (resultFeatureType != ResultFeatureType.PREDICTED_VALUE && resultFeatureType != ResultFeatureType.PREDICTED_DISPLAY_VALUE) {
fieldNames.add(getModelField(outputField));
}
});
return fieldNames;
}
use of org.dmg.pmml.OutputField in project jpmml-sparkml by jpmml.
the class GeneralizedLinearRegressionModelConverter method registerOutputFields.
@Override
public List<OutputField> registerOutputFields(Label label, SparkMLEncoder encoder) {
List<OutputField> result = super.registerOutputFields(label, encoder);
MiningFunction miningFunction = getMiningFunction();
switch(miningFunction) {
case CLASSIFICATION:
CategoricalLabel categoricalLabel = (CategoricalLabel) label;
result = new ArrayList<>(result);
result.addAll(ModelUtil.createProbabilityFields(DataType.DOUBLE, categoricalLabel.getValues()));
break;
default:
break;
}
return result;
}
Aggregations