Examples with TreeTargetNumericColumnData org.knime.base.node.mine.treeensemble.data.TreeTargetNumericColumnData used on opensource projects

Example 1 with TreeTargetNumericColumnData

use of org.knime.base.node.mine.treeensemble.data.TreeTargetNumericColumnData in project knime-core by knime.

the class TreeLearnerRegression method findBestSplitRegression.

private SplitCandidate findBestSplitRegression(final int currentDepth, final double[] rowSampleWeights, final TreeNodeSignature treeNodeSignature, final RegressionPriors targetPriors, final BitSet forbiddenColumnSet, final TreeNodeMembershipController membershipController) {
    final TreeData data = getData();
    final ColumnSampleStrategy colSamplingStrategy = getColSamplingStrategy();
    final TreeEnsembleLearnerConfiguration config = getConfig();
    final int maxLevels = config.getMaxLevels();
    if (maxLevels != TreeEnsembleLearnerConfiguration.MAX_LEVEL_INFINITE && currentDepth >= maxLevels) {
        return null;
    }
    final int minNodeSize = config.getMinNodeSize();
    if (minNodeSize != TreeEnsembleLearnerConfiguration.MIN_NODE_SIZE_UNDEFINED) {
        if (targetPriors.getNrRecords() < minNodeSize) {
            return null;
        }
    }
    final double priorSquaredDeviation = targetPriors.getSumSquaredDeviation();
    if (priorSquaredDeviation < TreeColumnData.EPSILON) {
        return null;
    }
    final TreeTargetNumericColumnData targetColumn = getTargetData();
    SplitCandidate splitCandidate = null;
    if (currentDepth == 0 && config.getHardCodedRootColumn() != null) {
        final TreeAttributeColumnData rootColumn = data.getColumn(config.getHardCodedRootColumn());
        return rootColumn.calcBestSplitRegression(membershipController, rowSampleWeights, targetPriors, targetColumn);
    } else {
        double bestGainValue = 0.0;
        final ColumnSample columnSample = colSamplingStrategy.getColumnSampleForTreeNode(treeNodeSignature);
        for (TreeAttributeColumnData col : columnSample) {
            if (forbiddenColumnSet.get(col.getMetaData().getAttributeIndex())) {
                continue;
            }
            SplitCandidate currentColSplit = col.calcBestSplitRegression(membershipController, rowSampleWeights, targetPriors, targetColumn);
            if (currentColSplit != null) {
                double gainValue = currentColSplit.getGainValue();
                if (gainValue > bestGainValue) {
                    bestGainValue = gainValue;
                    splitCandidate = currentColSplit;
                }
            }
        }
    }
    return splitCandidate;
}

Example 2 with TreeTargetNumericColumnData

use of org.knime.base.node.mine.treeensemble.data.TreeTargetNumericColumnData in project knime-core by knime.

the class TreeLearnerRegression method learnSingleTree.

/**
 * {@inheritDoc}
 */
@Override
public TreeModelRegression learnSingleTree(final ExecutionMonitor exec, final RandomData rd) throws CanceledExecutionException {
    final TreeTargetNumericColumnData targetColumn = getTargetData();
    final TreeData data = getData();
    final RowSample rowSampling = getRowSampling();
    final TreeEnsembleLearnerConfiguration config = getConfig();
    double[] dataMemberships = new double[data.getNrRows()];
    for (int i = 0; i < dataMemberships.length; i++) {
        dataMemberships[i] = rowSampling.getCountFor(i);
    }
    RegressionPriors targetPriors = targetColumn.getPriors(dataMemberships, config);
    BitSet forbiddenColumnSet = new BitSet(data.getNrAttributes());
    // TreeNodeMembershipController rootMembershipController = new TreeNodeMembershipController(data, dataMemberships);
    TreeNodeMembershipController rootMembershipController = null;
    TreeNodeRegression rootNode = buildTreeNode(exec, 0, dataMemberships, TreeNodeSignature.ROOT_SIGNATURE, targetPriors, forbiddenColumnSet, rootMembershipController);
    assert forbiddenColumnSet.cardinality() == 0;
    rootNode.setTreeNodeCondition(TreeNodeTrueCondition.INSTANCE);
    return new TreeModelRegression(rootNode);
}

Example 3 with TreeTargetNumericColumnData

use of org.knime.base.node.mine.treeensemble.data.TreeTargetNumericColumnData in project knime-core by knime.

the class TreeLearnerRegression method buildTreeNode.

private TreeNodeRegression buildTreeNode(final ExecutionMonitor exec, final int currentDepth, final double[] rowSampleWeights, final TreeNodeSignature treeNodeSignature, final RegressionPriors targetPriors, final BitSet forbiddenColumnSet, final TreeNodeMembershipController membershipController) throws CanceledExecutionException {
    final TreeData data = getData();
    final TreeEnsembleLearnerConfiguration config = getConfig();
    exec.checkCanceled();
    SplitCandidate bestSplit = findBestSplitRegression(currentDepth, rowSampleWeights, treeNodeSignature, targetPriors, forbiddenColumnSet, membershipController);
    if (bestSplit == null) {
        return new TreeNodeRegression(treeNodeSignature, targetPriors);
    }
    TreeAttributeColumnData splitColumn = bestSplit.getColumnData();
    final int attributeIndex = splitColumn.getMetaData().getAttributeIndex();
    boolean markAttributeAsForbidden = !bestSplit.canColumnBeSplitFurther();
    forbiddenColumnSet.set(attributeIndex, markAttributeAsForbidden);
    TreeNodeCondition[] childConditions = bestSplit.getChildConditions();
    if (childConditions.length > Short.MAX_VALUE) {
        throw new RuntimeException("Too many children when splitting " + "attribute " + bestSplit.getColumnData() + " (maximum supported: " + Short.MAX_VALUE + "): " + childConditions.length);
    }
    TreeNodeRegression[] childNodes = new TreeNodeRegression[childConditions.length];
    final double[] dataMemberships = rowSampleWeights;
    final double[] childMemberships = new double[dataMemberships.length];
    final TreeTargetNumericColumnData targetColumn = (TreeTargetNumericColumnData) data.getTargetColumn();
    for (int i = 0; i < childConditions.length; i++) {
        System.arraycopy(dataMemberships, 0, childMemberships, 0, dataMemberships.length);
        TreeNodeCondition cond = childConditions[i];
        splitColumn.updateChildMemberships(cond, dataMemberships, childMemberships);
        RegressionPriors childTargetPriors = targetColumn.getPriors(childMemberships, config);
        TreeNodeSignature childSignature = treeNodeSignature.createChildSignature((short) i);
        TreeNodeMembershipController childMembershipController = splitColumn.getChildNodeMembershipController(cond, membershipController);
        childNodes[i] = buildTreeNode(exec, currentDepth + 1, childMemberships, childSignature, childTargetPriors, forbiddenColumnSet, childMembershipController);
        childNodes[i].setTreeNodeCondition(cond);
    }
    if (markAttributeAsForbidden) {
        forbiddenColumnSet.set(attributeIndex, false);
    }
    return new TreeNodeRegression(treeNodeSignature, targetPriors, childNodes);
}