Search in sources :

Example 1 with PMMLDiscretize

use of org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize in project knime-core by knime.

the class AutoBinner method createDisretizeOp.

private PMMLPreprocDiscretize createDisretizeOp(final Map<String, double[]> edgesMap) {
    Map<String, List<PMMLDiscretizeBin>> binMap = createBins(edgesMap);
    List<String> names = new ArrayList<String>();
    Map<String, PMMLDiscretize> discretize = new HashMap<String, PMMLDiscretize>();
    for (String target : m_settings.getTargetColumn()) {
        String binnedCol = m_settings.getReplaceColumn() ? target : target + " [Binned]";
        names.add(binnedCol);
        discretize.put(binnedCol, new PMMLDiscretize(target, binMap.get(target)));
    }
    DisretizeConfiguration config = new DisretizeConfiguration(names, discretize);
    PMMLPreprocDiscretize op = new PMMLPreprocDiscretize(config);
    return op;
}
Also used : HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) DisretizeConfiguration(org.knime.base.node.preproc.autobinner.pmml.DisretizeConfiguration) ArrayList(java.util.ArrayList) PMMLDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize) ArrayList(java.util.ArrayList) List(java.util.List) PMMLPreprocDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLPreprocDiscretize)

Example 2 with PMMLDiscretize

use of org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize in project knime-core by knime.

the class DBAutoBinner 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<>();
    List<String> replacedColumnNames = pmmlDiscretize.getConfiguration().getNames();
    for (String replacedColumnName : replacedColumnNames) {
        PMMLDiscretize discretize = pmmlDiscretize.getConfiguration().getDiscretize(replacedColumnName);
        List<PMMLDiscretizeBin> bins = discretize.getBins();
        String originalColumnName = discretize.getField();
        if (replacedColumnName.equals(originalColumnName)) {
            // wenn replaced, dann nicht anhängen
            columnToAppend.put(originalColumnName, null);
        } else {
            // nicht replaced -> anhängen
            columnToAppend.put(originalColumnName, replacedColumnName);
        }
        NumericBin[] numericBin = new NumericBin[bins.size()];
        int counter = 0;
        for (PMMLDiscretizeBin bin : bins) {
            String binName = bin.getBinValue();
            List<PMMLInterval> intervals = bin.getIntervals();
            boolean leftOpen = false;
            boolean rightOpen = false;
            double leftMargin = 0;
            double rightMargin = 0;
            // always returns only one interval
            for (PMMLInterval interval : intervals) {
                Closure closure = interval.getClosure();
                switch(closure) {
                    case openClosed:
                        leftOpen = true;
                        rightOpen = false;
                        break;
                    case openOpen:
                        leftOpen = true;
                        rightOpen = true;
                        break;
                    case closedOpen:
                        leftOpen = false;
                        rightOpen = true;
                    case closedClosed:
                        leftOpen = false;
                        rightOpen = false;
                        break;
                    default:
                        leftOpen = true;
                        rightOpen = false;
                        break;
                }
                leftMargin = interval.getLeftMargin();
                rightMargin = interval.getRightMargin();
            }
            numericBin[counter] = new NumericBin(binName, leftOpen, leftMargin, rightOpen, rightMargin);
            counter++;
        }
        columnToBins.put(originalColumnName, numericBin);
    }
    // ColumnRearranger createColReg = createColReg(dataTableSpec, columnToBins, columnToAppended);
    DataTableSpec newDataTableSpec = createNewDataTableSpec(dataTableSpec, columnToAppend);
    PMMLPortObjectSpecCreator pmmlSpecCreator = new PMMLPortObjectSpecCreator(newDataTableSpec);
    PMMLPortObject pmmlPortObject = new PMMLPortObject(pmmlSpecCreator.createSpec(), null, newDataTableSpec);
    PMMLBinningTranslator trans = new PMMLBinningTranslator(columnToBins, columnToAppend, new DerivedFieldMapper(pmmlPortObject));
    TransformationDictionary exportToTransDict = trans.exportToTransDict();
    pmmlPortObject.addGlobalTransformations(exportToTransDict);
    return pmmlPortObject;
}
Also used : NumericBin(org.knime.base.node.preproc.pmml.binner.NumericBin) DataTableSpec(org.knime.core.data.DataTableSpec) Closure(org.knime.base.node.preproc.autobinner.pmml.PMMLInterval.Closure) HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) TransformationDictionary(org.dmg.pmml.TransformationDictionaryDocument.TransformationDictionary) PMMLDiscretizeBin(org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretizeBin) PMMLInterval(org.knime.base.node.preproc.autobinner.pmml.PMMLInterval) DerivedFieldMapper(org.knime.core.node.port.pmml.preproc.DerivedFieldMapper) PMMLPortObject(org.knime.core.node.port.pmml.PMMLPortObject) PMMLBinningTranslator(org.knime.base.node.preproc.pmml.binner.PMMLBinningTranslator) PMMLDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize) PMMLPortObjectSpecCreator(org.knime.core.node.port.pmml.PMMLPortObjectSpecCreator)

Example 3 with PMMLDiscretize

use of org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize in project knime-core by knime.

the class AutoBinner method createDisretizeOp.

private PMMLPreprocDiscretize createDisretizeOp(final Map<String, double[]> edgesMap) {
    Map<String, List<PMMLDiscretizeBin>> binMap = createBins(edgesMap);
    List<String> names = new ArrayList<String>();
    Map<String, PMMLDiscretize> discretize = new HashMap<String, PMMLDiscretize>();
    for (String target : m_settings.getTargetColumn()) {
        String binnedCol = m_settings.getReplaceColumn() ? target : target + " [Binned]";
        names.add(binnedCol);
        discretize.put(binnedCol, new PMMLDiscretize(target, binMap.get(target)));
    }
    DisretizeConfiguration config = new DisretizeConfiguration(names, discretize);
    PMMLPreprocDiscretize op = new PMMLPreprocDiscretize(config);
    return op;
}
Also used : HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) DisretizeConfiguration(org.knime.base.node.preproc.autobinner.pmml.DisretizeConfiguration) ArrayList(java.util.ArrayList) PMMLDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize) ArrayList(java.util.ArrayList) List(java.util.List) PMMLPreprocDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLPreprocDiscretize)

Example 4 with PMMLDiscretize

use of org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize in project knime-core by knime.

the class AutoBinner method createDisretizeOp.

/**
 * @param edgesMap the boundary map
 * @return the {@link PMMLPreprocDiscretize} model
 */
protected PMMLPreprocDiscretize createDisretizeOp(final Map<String, double[]> edgesMap) {
    Map<String, List<PMMLDiscretizeBin>> binMap = createBins(edgesMap);
    List<String> names = new ArrayList<String>();
    Map<String, PMMLDiscretize> discretize = new HashMap<String, PMMLDiscretize>();
    for (String target : m_included) {
        String binnedCol = m_settings.getReplaceColumn() ? target : target + " [Binned]";
        names.add(binnedCol);
        discretize.put(binnedCol, new PMMLDiscretize(target, binMap.get(target)));
    }
    DisretizeConfiguration config = new DisretizeConfiguration(names, discretize);
    PMMLPreprocDiscretize op = new PMMLPreprocDiscretize(config);
    return op;
}
Also used : HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) DisretizeConfiguration(org.knime.base.node.preproc.autobinner.pmml.DisretizeConfiguration) ArrayList(java.util.ArrayList) PMMLDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize) ArrayList(java.util.ArrayList) List(java.util.List) PMMLPreprocDiscretize(org.knime.base.node.preproc.autobinner.pmml.PMMLPreprocDiscretize)

Aggregations

HashMap (java.util.HashMap)4 LinkedHashMap (java.util.LinkedHashMap)4 PMMLDiscretize (org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretize)4 ArrayList (java.util.ArrayList)3 List (java.util.List)3 DisretizeConfiguration (org.knime.base.node.preproc.autobinner.pmml.DisretizeConfiguration)3 PMMLPreprocDiscretize (org.knime.base.node.preproc.autobinner.pmml.PMMLPreprocDiscretize)3 TransformationDictionary (org.dmg.pmml.TransformationDictionaryDocument.TransformationDictionary)1 PMMLDiscretizeBin (org.knime.base.node.preproc.autobinner.pmml.PMMLDiscretizeBin)1 PMMLInterval (org.knime.base.node.preproc.autobinner.pmml.PMMLInterval)1 Closure (org.knime.base.node.preproc.autobinner.pmml.PMMLInterval.Closure)1 NumericBin (org.knime.base.node.preproc.pmml.binner.NumericBin)1 PMMLBinningTranslator (org.knime.base.node.preproc.pmml.binner.PMMLBinningTranslator)1 DataTableSpec (org.knime.core.data.DataTableSpec)1 PMMLPortObject (org.knime.core.node.port.pmml.PMMLPortObject)1 PMMLPortObjectSpecCreator (org.knime.core.node.port.pmml.PMMLPortObjectSpecCreator)1 DerivedFieldMapper (org.knime.core.node.port.pmml.preproc.DerivedFieldMapper)1