use of org.apache.spark.ml.param.shared.HasProbabilityCol in project jpmml-sparkml by jpmml.
the class MultilayerPerceptronClassificationModelConverter method registerOutputFields.
@Override
public List<OutputField> registerOutputFields(Label label, SparkMLEncoder encoder) {
MultilayerPerceptronClassificationModel model = getTransformer();
List<OutputField> result = super.registerOutputFields(label, encoder);
if (!(model instanceof HasProbabilityCol)) {
CategoricalLabel categoricalLabel = (CategoricalLabel) label;
result = new ArrayList<>(result);
result.addAll(ModelUtil.createProbabilityFields(DataType.DOUBLE, categoricalLabel.getValues()));
}
return result;
}
use of org.apache.spark.ml.param.shared.HasProbabilityCol in project jpmml-sparkml by jpmml.
the class ClassificationModelConverter method registerOutputFields.
@Override
public List<OutputField> registerOutputFields(Label label, SparkMLEncoder encoder) {
T model = getTransformer();
CategoricalLabel categoricalLabel = (CategoricalLabel) label;
List<OutputField> result = new ArrayList<>();
String predictionCol = model.getPredictionCol();
OutputField pmmlPredictedField = ModelUtil.createPredictedField(FieldName.create("pmml(" + predictionCol + ")"), categoricalLabel.getDataType(), OpType.CATEGORICAL);
result.add(pmmlPredictedField);
List<String> categories = new ArrayList<>();
DocumentBuilder documentBuilder = DOMUtil.createDocumentBuilder();
InlineTable inlineTable = new InlineTable();
List<String> columns = Arrays.asList("input", "output");
for (int i = 0; i < categoricalLabel.size(); i++) {
String value = categoricalLabel.getValue(i);
String category = String.valueOf(i);
categories.add(category);
Row row = DOMUtil.createRow(documentBuilder, columns, Arrays.asList(value, category));
inlineTable.addRows(row);
}
MapValues mapValues = new MapValues().addFieldColumnPairs(new FieldColumnPair(pmmlPredictedField.getName(), columns.get(0))).setOutputColumn(columns.get(1)).setInlineTable(inlineTable);
final OutputField predictedField = new OutputField(FieldName.create(predictionCol), DataType.DOUBLE).setOpType(OpType.CATEGORICAL).setResultFeature(ResultFeature.TRANSFORMED_VALUE).setExpression(mapValues);
result.add(predictedField);
Feature feature = new CategoricalFeature(encoder, predictedField.getName(), predictedField.getDataType(), categories) {
@Override
public ContinuousFeature toContinuousFeature() {
PMMLEncoder encoder = ensureEncoder();
return new ContinuousFeature(encoder, getName(), getDataType());
}
};
encoder.putOnlyFeature(predictionCol, feature);
if (model instanceof HasProbabilityCol) {
HasProbabilityCol hasProbabilityCol = (HasProbabilityCol) model;
String probabilityCol = hasProbabilityCol.getProbabilityCol();
List<Feature> features = new ArrayList<>();
for (int i = 0; i < categoricalLabel.size(); i++) {
String value = categoricalLabel.getValue(i);
OutputField probabilityField = ModelUtil.createProbabilityField(FieldName.create(probabilityCol + "(" + value + ")"), DataType.DOUBLE, value);
result.add(probabilityField);
features.add(new ContinuousFeature(encoder, probabilityField.getName(), probabilityField.getDataType()));
}
encoder.putFeatures(probabilityCol, features);
}
return result;
}
Aggregations