Search in sources :

Example 6 with MiningField

use of org.dmg.pmml.MiningField in project drools by kiegroup.

the class ModelUtilsTest method getTargetFieldsTypeMapWithTargetFieldsWithoutTargets.

@Test
public void getTargetFieldsTypeMapWithTargetFieldsWithoutTargets() {
    final Model model = new RegressionModel();
    final DataDictionary dataDictionary = new DataDictionary();
    final MiningSchema miningSchema = new MiningSchema();
    IntStream.range(0, 3).forEach(i -> {
        final DataField dataField = getRandomDataField();
        dataDictionary.addDataFields(dataField);
        final MiningField miningField = getMiningField(dataField.getName().getValue(), MiningField.UsageType.PREDICTED);
        miningSchema.addMiningFields(miningField);
    });
    model.setMiningSchema(miningSchema);
    Map<String, DATA_TYPE> retrieved = ModelUtils.getTargetFieldsTypeMap(getFieldsFromDataDictionary(dataDictionary), model);
    assertNotNull(retrieved);
    assertEquals(miningSchema.getMiningFields().size(), retrieved.size());
    assertTrue(retrieved instanceof LinkedHashMap);
    final Iterator<Map.Entry<String, DATA_TYPE>> iterator = retrieved.entrySet().iterator();
    for (int i = 0; i < miningSchema.getMiningFields().size(); i++) {
        MiningField miningField = miningSchema.getMiningFields().get(i);
        DataField dataField = dataDictionary.getDataFields().stream().filter(df -> df.getName().equals(miningField.getName())).findFirst().get();
        DATA_TYPE expected = DATA_TYPE.byName(dataField.getDataType().value());
        final Map.Entry<String, DATA_TYPE> next = iterator.next();
        assertEquals(expected, next.getValue());
    }
}
Also used : PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) MiningField(org.dmg.pmml.MiningField) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) DataDictionary(org.dmg.pmml.DataDictionary) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) RegressionModel(org.dmg.pmml.regression.RegressionModel) LinkedHashMap(java.util.LinkedHashMap) MiningSchema(org.dmg.pmml.MiningSchema) DataField(org.dmg.pmml.DataField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) Model(org.dmg.pmml.Model) RegressionModel(org.dmg.pmml.regression.RegressionModel) DATA_TYPE(org.kie.pmml.api.enums.DATA_TYPE) Map(java.util.Map) HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) Test(org.junit.Test)

Example 7 with MiningField

use of org.dmg.pmml.MiningField in project drools by kiegroup.

the class ModelUtilsTest method convertToKieMiningField.

@Test
public void convertToKieMiningField() {
    final String fieldName = "fieldName";
    final MiningField.UsageType usageType = MiningField.UsageType.ACTIVE;
    final MiningField toConvert = getMiningField(fieldName, usageType);
    toConvert.setOpType(null);
    final DataField dataField = getDataField(fieldName, OpType.CATEGORICAL, DataType.STRING);
    org.kie.pmml.api.models.MiningField retrieved = ModelUtils.convertToKieMiningField(toConvert, dataField);
    assertNotNull(retrieved);
    assertEquals(fieldName, retrieved.getName());
    assertEquals(FIELD_USAGE_TYPE.ACTIVE, retrieved.getUsageType());
    assertEquals(DATA_TYPE.STRING, retrieved.getDataType());
    assertNull(retrieved.getOpType());
    toConvert.setOpType(OpType.CATEGORICAL);
    retrieved = ModelUtils.convertToKieMiningField(toConvert, dataField);
    assertEquals(OP_TYPE.CATEGORICAL, retrieved.getOpType());
}
Also used : PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) MiningField(org.dmg.pmml.MiningField) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) DataField(org.dmg.pmml.DataField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) Test(org.junit.Test)

Example 8 with MiningField

use of org.dmg.pmml.MiningField in project drools by kiegroup.

the class ModelUtilsTest method getTargetFieldTypeWithoutTargetField.

@Test(expected = Exception.class)
public void getTargetFieldTypeWithoutTargetField() {
    final String fieldName = "fieldName";
    MiningField.UsageType usageType = MiningField.UsageType.ACTIVE;
    MiningField miningField = getMiningField(fieldName, usageType);
    final DataField dataField = getDataField(fieldName, OpType.CATEGORICAL, DataType.STRING);
    final DataDictionary dataDictionary = new DataDictionary();
    dataDictionary.addDataFields(dataField);
    final MiningSchema miningSchema = new MiningSchema();
    miningSchema.addMiningFields(miningField);
    final Model model = new RegressionModel();
    model.setMiningSchema(miningSchema);
    ModelUtils.getTargetFieldType(getFieldsFromDataDictionary(dataDictionary), model);
}
Also used : PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) MiningField(org.dmg.pmml.MiningField) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) DataField(org.dmg.pmml.DataField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) MiningSchema(org.dmg.pmml.MiningSchema) Model(org.dmg.pmml.Model) RegressionModel(org.dmg.pmml.regression.RegressionModel) DataDictionary(org.dmg.pmml.DataDictionary) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) RegressionModel(org.dmg.pmml.regression.RegressionModel) Test(org.junit.Test)

Example 9 with MiningField

use of org.dmg.pmml.MiningField in project drools by kiegroup.

the class ModelUtilsTest method getTargetFieldsWithTargetFieldsWithTargetsWithoutOptType.

@Test
public void getTargetFieldsWithTargetFieldsWithTargetsWithoutOptType() {
    final Model model = new RegressionModel();
    final DataDictionary dataDictionary = new DataDictionary();
    final MiningSchema miningSchema = new MiningSchema();
    final Targets targets = new Targets();
    IntStream.range(0, 3).forEach(i -> {
        final String fieldName = "fieldName-" + i;
        final DataField dataField = getDataField(fieldName, OpType.CATEGORICAL, DataType.STRING);
        dataDictionary.addDataFields(dataField);
        final MiningField miningField = getMiningField(fieldName, MiningField.UsageType.PREDICTED);
        miningField.setOpType(OpType.CONTINUOUS);
        miningSchema.addMiningFields(miningField);
        final Target targetField = getTarget(fieldName, null);
        targets.addTargets(targetField);
    });
    model.setMiningSchema(miningSchema);
    model.setTargets(targets);
    List<KiePMMLNameOpType> retrieved = ModelUtils.getTargetFields(getFieldsFromDataDictionary(dataDictionary), model);
    assertNotNull(retrieved);
    assertEquals(miningSchema.getMiningFields().size(), retrieved.size());
    retrieved.forEach(kiePMMLNameOpType -> {
        Optional<MiningField> optionalMiningField = miningSchema.getMiningFields().stream().filter(fld -> kiePMMLNameOpType.getName().equals(fld.getName().getValue())).findFirst();
        assertTrue(optionalMiningField.isPresent());
        MiningField miningField = optionalMiningField.get();
        OP_TYPE expected = OP_TYPE.byName(miningField.getOpType().value());
        assertEquals(expected, kiePMMLNameOpType.getOpType());
    });
}
Also used : RESULT_FEATURE(org.kie.pmml.api.enums.RESULT_FEATURE) ModelUtils.getPrefixedName(org.kie.pmml.compiler.api.utils.ModelUtils.getPrefixedName) Arrays(java.util.Arrays) Date(java.util.Date) Model(org.dmg.pmml.Model) MiningSchema(org.dmg.pmml.MiningSchema) OP_TYPE(org.kie.pmml.api.enums.OP_TYPE) Row(org.dmg.pmml.Row) FieldName(org.dmg.pmml.FieldName) FIELD_USAGE_TYPE(org.kie.pmml.api.enums.FIELD_USAGE_TYPE) OpType(org.dmg.pmml.OpType) Map(java.util.Map) InputCell(org.jpmml.model.inlinetable.InputCell) PMMLModelTestUtils.getDataTypes(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataTypes) PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) RegressionModel(org.dmg.pmml.regression.RegressionModel) Targets(org.dmg.pmml.Targets) DataType(org.dmg.pmml.DataType) PMMLModelTestUtils.getRandomRow(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomRow) Collectors(java.util.stream.Collectors) PMMLModelTestUtils.getParameterFields(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getParameterFields) DataField(org.dmg.pmml.DataField) List(java.util.List) Assert.assertFalse(org.junit.Assert.assertFalse) Optional(java.util.Optional) PMMLModelTestUtils.getArray(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getArray) PMMLModelTestUtils.getRandomDataType(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataType) ParameterField(org.dmg.pmml.ParameterField) IntStream(java.util.stream.IntStream) OutputField(org.dmg.pmml.OutputField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) HashMap(java.util.HashMap) ArrayList(java.util.ArrayList) LinkedHashMap(java.util.LinkedHashMap) Field(org.dmg.pmml.Field) DerivedField(org.dmg.pmml.DerivedField) KiePMMLInternalException(org.kie.pmml.api.exceptions.KiePMMLInternalException) KiePMMLNameOpType(org.kie.pmml.commons.model.tuples.KiePMMLNameOpType) OutputCell(org.jpmml.model.inlinetable.OutputCell) MiningField(org.dmg.pmml.MiningField) Iterator(java.util.Iterator) Assert.assertNotNull(org.junit.Assert.assertNotNull) Assert.assertTrue(org.junit.Assert.assertTrue) DataDictionary(org.dmg.pmml.DataDictionary) Test(org.junit.Test) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) Target(org.dmg.pmml.Target) DATA_TYPE(org.kie.pmml.api.enums.DATA_TYPE) Array(org.dmg.pmml.Array) PMMLModelTestUtils.getRandomOutputField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomOutputField) PMMLModelTestUtils.getRandomTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomTarget) Assert.assertNull(org.junit.Assert.assertNull) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) PMMLModelTestUtils.getTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getTarget) PMMLModelTestUtils.getRandomRowWithCells(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomRowWithCells) Assert.assertEquals(org.junit.Assert.assertEquals) PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) MiningField(org.dmg.pmml.MiningField) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) Targets(org.dmg.pmml.Targets) OP_TYPE(org.kie.pmml.api.enums.OP_TYPE) DataDictionary(org.dmg.pmml.DataDictionary) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) RegressionModel(org.dmg.pmml.regression.RegressionModel) KiePMMLNameOpType(org.kie.pmml.commons.model.tuples.KiePMMLNameOpType) Target(org.dmg.pmml.Target) PMMLModelTestUtils.getRandomTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomTarget) PMMLModelTestUtils.getTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getTarget) MiningSchema(org.dmg.pmml.MiningSchema) DataField(org.dmg.pmml.DataField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) Model(org.dmg.pmml.Model) RegressionModel(org.dmg.pmml.regression.RegressionModel) Test(org.junit.Test)

Example 10 with MiningField

use of org.dmg.pmml.MiningField in project drools by kiegroup.

the class ModelUtilsTest method getTargetFieldsWithTargetFieldsWithOptType.

@Test
public void getTargetFieldsWithTargetFieldsWithOptType() {
    final Model model = new RegressionModel();
    final DataDictionary dataDictionary = new DataDictionary();
    final MiningSchema miningSchema = new MiningSchema();
    IntStream.range(0, 3).forEach(i -> {
        final String fieldName = "fieldName-" + i;
        final DataField dataField = getDataField(fieldName, OpType.CATEGORICAL, DataType.STRING);
        dataDictionary.addDataFields(dataField);
        final MiningField miningField = getMiningField(fieldName, MiningField.UsageType.PREDICTED);
        miningField.setOpType(OpType.CONTINUOUS);
        miningSchema.addMiningFields(miningField);
    });
    model.setMiningSchema(miningSchema);
    List<KiePMMLNameOpType> retrieved = ModelUtils.getTargetFields(getFieldsFromDataDictionary(dataDictionary), model);
    assertNotNull(retrieved);
    assertEquals(miningSchema.getMiningFields().size(), retrieved.size());
    retrieved.forEach(kiePMMLNameOpType -> {
        Optional<MiningField> optionalMiningField = miningSchema.getMiningFields().stream().filter(fld -> kiePMMLNameOpType.getName().equals(fld.getName().getValue())).findFirst();
        assertTrue(optionalMiningField.isPresent());
        MiningField miningField = optionalMiningField.get();
        OP_TYPE expected = OP_TYPE.byName(miningField.getOpType().value());
        assertEquals(expected, kiePMMLNameOpType.getOpType());
    });
}
Also used : KiePMMLNameOpType(org.kie.pmml.commons.model.tuples.KiePMMLNameOpType) RESULT_FEATURE(org.kie.pmml.api.enums.RESULT_FEATURE) ModelUtils.getPrefixedName(org.kie.pmml.compiler.api.utils.ModelUtils.getPrefixedName) Arrays(java.util.Arrays) Date(java.util.Date) Model(org.dmg.pmml.Model) MiningSchema(org.dmg.pmml.MiningSchema) OP_TYPE(org.kie.pmml.api.enums.OP_TYPE) Row(org.dmg.pmml.Row) FieldName(org.dmg.pmml.FieldName) FIELD_USAGE_TYPE(org.kie.pmml.api.enums.FIELD_USAGE_TYPE) OpType(org.dmg.pmml.OpType) Map(java.util.Map) InputCell(org.jpmml.model.inlinetable.InputCell) PMMLModelTestUtils.getDataTypes(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataTypes) PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) RegressionModel(org.dmg.pmml.regression.RegressionModel) Targets(org.dmg.pmml.Targets) DataType(org.dmg.pmml.DataType) PMMLModelTestUtils.getRandomRow(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomRow) Collectors(java.util.stream.Collectors) PMMLModelTestUtils.getParameterFields(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getParameterFields) DataField(org.dmg.pmml.DataField) List(java.util.List) Assert.assertFalse(org.junit.Assert.assertFalse) Optional(java.util.Optional) PMMLModelTestUtils.getArray(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getArray) PMMLModelTestUtils.getRandomDataType(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataType) ParameterField(org.dmg.pmml.ParameterField) IntStream(java.util.stream.IntStream) OutputField(org.dmg.pmml.OutputField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) HashMap(java.util.HashMap) ArrayList(java.util.ArrayList) LinkedHashMap(java.util.LinkedHashMap) Field(org.dmg.pmml.Field) DerivedField(org.dmg.pmml.DerivedField) KiePMMLInternalException(org.kie.pmml.api.exceptions.KiePMMLInternalException) KiePMMLNameOpType(org.kie.pmml.commons.model.tuples.KiePMMLNameOpType) OutputCell(org.jpmml.model.inlinetable.OutputCell) MiningField(org.dmg.pmml.MiningField) Iterator(java.util.Iterator) Assert.assertNotNull(org.junit.Assert.assertNotNull) Assert.assertTrue(org.junit.Assert.assertTrue) DataDictionary(org.dmg.pmml.DataDictionary) Test(org.junit.Test) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) Target(org.dmg.pmml.Target) DATA_TYPE(org.kie.pmml.api.enums.DATA_TYPE) Array(org.dmg.pmml.Array) PMMLModelTestUtils.getRandomOutputField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomOutputField) PMMLModelTestUtils.getRandomTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomTarget) Assert.assertNull(org.junit.Assert.assertNull) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) PMMLModelTestUtils.getTarget(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getTarget) PMMLModelTestUtils.getRandomRowWithCells(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomRowWithCells) Assert.assertEquals(org.junit.Assert.assertEquals) PMMLModelTestUtils.getMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField) MiningField(org.dmg.pmml.MiningField) PMMLModelTestUtils.getRandomMiningField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField) MiningSchema(org.dmg.pmml.MiningSchema) DataField(org.dmg.pmml.DataField) PMMLModelTestUtils.getDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField) PMMLModelTestUtils.getRandomDataField(org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField) Model(org.dmg.pmml.Model) RegressionModel(org.dmg.pmml.regression.RegressionModel) OP_TYPE(org.kie.pmml.api.enums.OP_TYPE) DataDictionary(org.dmg.pmml.DataDictionary) CommonTestingUtils.getFieldsFromDataDictionary(org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary) RegressionModel(org.dmg.pmml.regression.RegressionModel) Test(org.junit.Test)

Aggregations

MiningField (org.dmg.pmml.MiningField)59 DataField (org.dmg.pmml.DataField)40 Test (org.junit.Test)39 MiningSchema (org.dmg.pmml.MiningSchema)33 DataDictionary (org.dmg.pmml.DataDictionary)25 RegressionModel (org.dmg.pmml.regression.RegressionModel)24 Model (org.dmg.pmml.Model)22 PMMLModelTestUtils.getRandomDataField (org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomDataField)20 PMML (org.dmg.pmml.PMML)18 PMMLModelTestUtils.getRandomMiningField (org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getRandomMiningField)18 PMMLModelTestUtils.getMiningField (org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getMiningField)17 PMMLModelTestUtils.getDataField (org.kie.pmml.compiler.api.testutils.PMMLModelTestUtils.getDataField)16 OutputField (org.dmg.pmml.OutputField)15 CommonTestingUtils.getFieldsFromDataDictionary (org.kie.pmml.compiler.api.CommonTestingUtils.getFieldsFromDataDictionary)15 FieldName (org.dmg.pmml.FieldName)12 Target (org.dmg.pmml.Target)11 Targets (org.dmg.pmml.Targets)11 OP_TYPE (org.kie.pmml.api.enums.OP_TYPE)11 HashMap (java.util.HashMap)10 List (java.util.List)10