Examples with PMMLDiscretize org.knime.core.util.binning.auto.pmml.PMMLDiscretize used on opensource projects

Example 1 with PMMLDiscretize

use of org.knime.core.util.binning.auto.pmml.PMMLDiscretize in project knime-core by knime.

the class AutoBinnerUtility method createDisretizeOp.

private static PMMLPreprocDiscretize createDisretizeOp(final AutoBinnerLearnSettings settings, final Map<String, double[]> edgesMap, final String[] includeCols) {
    final Map<String, List<PMMLDiscretizeBin>> binMap = createBins(settings, edgesMap, includeCols);
    final List<String> names = new ArrayList<String>();
    final Map<String, PMMLDiscretize> discretize = new HashMap<String, PMMLDiscretize>();
    for (String target : includeCols) {
        final String binnedCol = settings.getReplaceColumn() ? target : target + " [Binned]";
        names.add(binnedCol);
        discretize.put(binnedCol, new PMMLDiscretize(target, binMap.get(target)));
    }
    final DisretizeConfiguration config = new DisretizeConfiguration(names, discretize);
    return new PMMLPreprocDiscretize(config);
}

Example 2 with PMMLDiscretize

use of org.knime.core.util.binning.auto.pmml.PMMLDiscretize in project knime-core by knime.

the class AutoBinnerUtility method translate.

/**
 * This method translates a {@link PMMLPreprocDiscretize} object into {@link PMMLPortObject}.
 *
 * @param pmmlDiscretize {@link PMMLPreprocDiscretize} object
 * @param dataTableSpec {@link DataTableSpec} if incoming {@link BufferedDataTable}
 * @return a {@link PMMLPortObject} containing required parameters for binning operation
 */
public static PMMLPortObject translate(final PMMLPreprocDiscretize pmmlDiscretize, final DataTableSpec dataTableSpec) {
    final Map<String, Bin[]> columnToBins = new HashMap<>();
    final Map<String, String> columnToAppend = new HashMap<>();
    final List<String> replacedColumnNames = pmmlDiscretize.getConfiguration().getNames();
    for (String replacedColumnName : replacedColumnNames) {
        final PMMLDiscretize discretize = pmmlDiscretize.getConfiguration().getDiscretize(replacedColumnName);
        final List<PMMLDiscretizeBin> bins = discretize.getBins();
        final String originalColumnName = discretize.getField();
        final boolean replaceColumnTheSame = replacedColumnName.equals(originalColumnName);
        columnToAppend.put(originalColumnName, replaceColumnTheSame ? null : replacedColumnName);
        final NumericBin[] numericBin = bins.stream().map(AutoBinnerUtility::getNumericBin).toArray(NumericBin[]::new);
        columnToBins.put(originalColumnName, numericBin);
    }
    final DataTableSpec newDataTableSpec = createNewDataTableSpec(dataTableSpec, columnToAppend);
    final PMMLPortObjectSpecCreator pmmlSpecCreator = new PMMLPortObjectSpecCreator(newDataTableSpec);
    final PMMLPortObject pmmlPortObject = new PMMLPortObject(pmmlSpecCreator.createSpec(), null, newDataTableSpec);
    final PMMLBinningTranslator trans = new PMMLBinningTranslator(columnToBins, columnToAppend, new DerivedFieldMapper(pmmlPortObject));
    final TransformationDictionary exportToTransDict = trans.exportToTransDict();
    pmmlPortObject.addGlobalTransformations(exportToTransDict);
    return pmmlPortObject;
}