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

Example 1 with NominalValueRepresentation

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

the class NominalSplitCandidate method getChildConditions.

/**
 * {@inheritDoc}
 */
@Override
public TreeNodeCondition[] getChildConditions() {
    TreeNominalColumnMetaData columnMeta = getColumnData().getMetaData();
    NominalValueRepresentation[] values = columnMeta.getValues();
    List<TreeNodeCondition> resultList = new ArrayList<TreeNodeCondition>(values.length);
    for (int i = 0; i < values.length; i++) {
        if (m_sumWeightsAttributes[i] >= TreeColumnData.EPSILON) {
            resultList.add(new TreeNodeNominalCondition(columnMeta, i));
        }
    }
    return resultList.toArray(new TreeNodeCondition[resultList.size()]);
}

Example 2 with NominalValueRepresentation

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

the class TreeNodeClassification method createDecisionTreeNode.

/**
 * Creates DecisionTreeNode model that is used in Decision Tree of KNIME
 *
 * @param idGenerator
 * @param metaData
 * @return a DecisionTreeNode
 */
public DecisionTreeNode createDecisionTreeNode(final MutableInteger idGenerator, final TreeMetaData metaData) {
    DataCell majorityCell = new StringCell(getMajorityClassName());
    double[] targetDistribution = getTargetDistribution();
    int initSize = (int) (targetDistribution.length / 0.75 + 1.0);
    LinkedHashMap<DataCell, Double> scoreDistributionMap = new LinkedHashMap<DataCell, Double>(initSize);
    NominalValueRepresentation[] targets = getTargetMetaData().getValues();
    for (int i = 0; i < targetDistribution.length; i++) {
        String cl = targets[i].getNominalValue();
        double d = targetDistribution[i];
        scoreDistributionMap.put(new StringCell(cl), d);
    }
    final int nrChildren = getNrChildren();
    if (nrChildren == 0) {
        return new DecisionTreeNodeLeaf(idGenerator.inc(), majorityCell, scoreDistributionMap);
    } else {
        int id = idGenerator.inc();
        DecisionTreeNode[] childNodes = new DecisionTreeNode[nrChildren];
        int splitAttributeIndex = getSplitAttributeIndex();
        assert splitAttributeIndex >= 0 : "non-leaf node has no split";
        String splitAttribute = metaData.getAttributeMetaData(splitAttributeIndex).getAttributeName();
        PMMLPredicate[] childPredicates = new PMMLPredicate[nrChildren];
        for (int i = 0; i < nrChildren; i++) {
            final TreeNodeClassification treeNode = getChild(i);
            TreeNodeCondition cond = treeNode.getCondition();
            childPredicates[i] = cond.toPMMLPredicate();
            childNodes[i] = treeNode.createDecisionTreeNode(idGenerator, metaData);
        }
        return new DecisionTreeNodeSplitPMML(id, majorityCell, scoreDistributionMap, splitAttribute, childPredicates, childNodes);
    }
}

Example 3 with NominalValueRepresentation

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

the class TreeModelPMMLTranslator method addTreeNode.

/**
 * @param pmmlNode
 * @param node
 */
private void addTreeNode(final Node pmmlNode, final TreeNodeClassification node) {
    int index = m_nodeIndex++;
    pmmlNode.setId(Integer.toString(index));
    pmmlNode.setScore(node.getMajorityClassName());
    double[] targetDistribution = node.getTargetDistribution();
    NominalValueRepresentation[] targetVals = node.getTargetMetaData().getValues();
    double sum = 0.0;
    for (Double v : targetDistribution) {
        sum += v;
    }
    pmmlNode.setRecordCount(sum);
    TreeNodeCondition condition = node.getCondition();
    if (condition instanceof TreeNodeTrueCondition) {
        pmmlNode.addNewTrue();
    } else if (condition instanceof TreeNodeColumnCondition) {
        final TreeNodeColumnCondition colCondition = (TreeNodeColumnCondition) condition;
        final String colName = colCondition.getColumnMetaData().getAttributeName();
        final Operator.Enum operator;
        final String value;
        if (condition instanceof TreeNodeNominalCondition) {
            final TreeNodeNominalCondition nominalCondition = (TreeNodeNominalCondition) condition;
            operator = Operator.EQUAL;
            value = nominalCondition.getValue();
        } else if (condition instanceof TreeNodeBitCondition) {
            final TreeNodeBitCondition bitCondition = (TreeNodeBitCondition) condition;
            operator = Operator.EQUAL;
            value = bitCondition.getValue() ? "1" : "0";
        } else if (condition instanceof TreeNodeNumericCondition) {
            final TreeNodeNumericCondition numCondition = (TreeNodeNumericCondition) condition;
            NumericOperator numOperator = numCondition.getNumericOperator();
            switch(numOperator) {
                case LargerThan:
                    operator = Operator.GREATER_THAN;
                    break;
                case LessThanOrEqual:
                    operator = Operator.LESS_OR_EQUAL;
                    break;
                default:
                    throw new IllegalStateException("Unsupported operator (not " + "implemented): " + numOperator);
            }
            value = Double.toString(numCondition.getSplitValue());
        } else {
            throw new IllegalStateException("Unsupported condition (not " + "implemented): " + condition.getClass().getSimpleName());
        }
        SimplePredicate pmmlSimplePredicate = pmmlNode.addNewSimplePredicate();
        pmmlSimplePredicate.setField(colName);
        pmmlSimplePredicate.setOperator(operator);
        pmmlSimplePredicate.setValue(value);
    } else {
        throw new IllegalStateException("Unsupported condition (not " + "implemented): " + condition.getClass().getSimpleName());
    }
    // adding score distribution (class counts)
    for (int i = 0; i < targetDistribution.length; i++) {
        String className = targetVals[i].getNominalValue();
        double freq = targetDistribution[i];
        ScoreDistribution pmmlScoreDist = pmmlNode.addNewScoreDistribution();
        pmmlScoreDist.setValue(className);
        pmmlScoreDist.setRecordCount(freq);
    }
    for (int i = 0; i < node.getNrChildren(); i++) {
        addTreeNode(pmmlNode.addNewNode(), node.getChild(i));
    }
}