Examples with DoubleValue org.knime.core.data.DoubleValue used on opensource projects

Example 51 with DoubleValue

use of org.knime.core.data.DoubleValue in project knime-core by knime.

the class FuzzyClusterNodeModel method execute.

/**
 * Generate new clustering based on InputDataTable and specified number of
 * clusters. In the output table, you will find the datarow with
 * supplementary information about the membership to each cluster center.
 * OUTPORT = original datarows with cluster membership information
 *
 * {@inheritDoc}
 */
@Override
protected PortObject[] execute(final PortObject[] inData, final ExecutionContext exec) throws Exception {
    BufferedDataTable indata = (BufferedDataTable) inData[0];
    m_clusters = null;
    m_betweenClusterVariation = Double.NaN;
    m_withinClusterVariation = null;
    if (m_noise) {
        if (m_calculateDelta) {
            if (m_memory) {
                m_fcmAlgo = new FCMAlgorithmMemory(m_nrClusters, m_fuzzifier, m_calculateDelta, m_lambda);
            } else {
                m_fcmAlgo = new FCMAlgorithm(m_nrClusters, m_fuzzifier, m_calculateDelta, m_lambda);
            }
        } else {
            if (m_memory) {
                m_fcmAlgo = new FCMAlgorithmMemory(m_nrClusters, m_fuzzifier, m_calculateDelta, m_delta);
            } else {
                m_fcmAlgo = new FCMAlgorithm(m_nrClusters, m_fuzzifier, m_calculateDelta, m_delta);
            }
        }
    } else {
        if (m_memory) {
            m_fcmAlgo = new FCMAlgorithmMemory(m_nrClusters, m_fuzzifier);
        } else {
            m_fcmAlgo = new FCMAlgorithm(m_nrClusters, m_fuzzifier);
        }
    }
    int nrRows = indata.getRowCount();
    DataTableSpec spec = indata.getDataTableSpec();
    int nrCols = spec.getNumColumns();
    List<String> learningCols = new LinkedList<String>();
    List<String> ignoreCols = new LinkedList<String>();
    // counter for included columns
    int z = 0;
    final int[] columns = new int[m_list.size()];
    for (int i = 0; i < nrCols; i++) {
        // if include does contain current column name
        String colname = spec.getColumnSpec(i).getName();
        if (m_list.contains(colname)) {
            columns[z] = i;
            z++;
            learningCols.add(colname);
        } else {
            ignoreCols.add(colname);
        }
    }
    ColumnRearranger colre = new ColumnRearranger(spec);
    colre.keepOnly(columns);
    BufferedDataTable filteredtable = exec.createColumnRearrangeTable(indata, colre, exec);
    // get dimension of feature space
    int dimension = filteredtable.getDataTableSpec().getNumColumns();
    Random random = new Random();
    if (m_useRandomSeed) {
        random.setSeed(m_randomSeed);
    }
    m_fcmAlgo.init(nrRows, dimension, filteredtable, random);
    // main loop - until clusters stop changing or maxNrIterations reached
    int currentIteration = 0;
    double totalchange = Double.MAX_VALUE;
    while ((totalchange > 1e-7) && (currentIteration < m_maxNrIterations)) {
        exec.checkCanceled();
        exec.setProgress((double) currentIteration / (double) m_maxNrIterations, "Iteration " + currentIteration + " Total change of prototypes: " + totalchange);
        totalchange = m_fcmAlgo.doOneIteration(exec);
        currentIteration++;
    }
    if (m_measures) {
        double[][] data = null;
        if (m_fcmAlgo instanceof FCMAlgorithmMemory) {
            data = ((FCMAlgorithmMemory) m_fcmAlgo).getConvertedData();
        } else {
            data = new double[nrRows][m_fcmAlgo.getDimension()];
            int curRow = 0;
            for (DataRow dRow : filteredtable) {
                for (int j = 0; j < dRow.getNumCells(); j++) {
                    if (!(dRow.getCell(j).isMissing())) {
                        DoubleValue dv = (DoubleValue) dRow.getCell(j);
                        data[curRow][j] = dv.getDoubleValue();
                    } else {
                        data[curRow][j] = 0;
                    }
                }
                curRow++;
            }
        }
        m_fcmmeasures = new FCMQualityMeasures(m_fcmAlgo.getClusterCentres(), m_fcmAlgo.getweightMatrix(), data, m_fuzzifier);
    }
    ColumnRearranger colRearranger = new ColumnRearranger(spec);
    CellFactory membershipFac = new ClusterMembershipFactory(m_fcmAlgo);
    colRearranger.append(membershipFac);
    BufferedDataTable result = exec.createColumnRearrangeTable(indata, colRearranger, exec);
    // don't write out the noise cluster!
    double[][] clustercentres = m_fcmAlgo.getClusterCentres();
    if (m_noise) {
        double[][] cleaned = new double[clustercentres.length - 1][];
        for (int i = 0; i < cleaned.length; i++) {
            cleaned[i] = new double[clustercentres[i].length];
            System.arraycopy(clustercentres[i], 0, cleaned[i], 0, clustercentres[i].length);
        }
        clustercentres = cleaned;
    }
    exec.setMessage("Creating PMML cluster model...");
    // handle the optional PMML input
    PMMLPortObject inPMMLPort = m_enablePMMLInput ? (PMMLPortObject) inData[1] : null;
    PMMLPortObjectSpec inPMMLSpec = null;
    if (inPMMLPort != null) {
        inPMMLSpec = inPMMLPort.getSpec();
    }
    PMMLPortObjectSpec pmmlOutSpec = createPMMLPortObjectSpec(inPMMLSpec, spec, learningCols);
    PMMLPortObject outPMMLPort = new PMMLPortObject(pmmlOutSpec, inPMMLPort, spec);
    outPMMLPort.addModelTranslater(new PMMLClusterTranslator(ComparisonMeasure.squaredEuclidean, m_nrClusters, clustercentres, null, new LinkedHashSet<String>(pmmlOutSpec.getLearningFields())));
    return new PortObject[] { result, outPMMLPort };
}

Example 52 with DoubleValue

use of org.knime.core.data.DoubleValue in project knime-core by knime.

the class MDSProjectionManager method preprocFixedDataPoints.

/**
 * Initializes for each of the fixed data points a point in the
 * target space. Which of the columns of the data table containing the
 * fixed points have to be considered (according to the non fixed points)
 * is specified by the given array of indices.
 *
 * @param fixedDataMdsIndices The indices specifying the columns of
 * the data table containing the fixed data points, to consider.
 * @throws CanceledExecutionException If the process is canceled.
 */
protected void preprocFixedDataPoints(final int[] fixedDataMdsIndices) throws CanceledExecutionException {
    m_exec.setMessage("Preprocessing fixed data points");
    // sort indices
    Arrays.sort(fixedDataMdsIndices);
    RowIterator it = m_fixedDataPoints.iterator();
    while (it.hasNext()) {
        m_exec.checkCanceled();
        DataRow row = it.next();
        DataPoint p = new DataPoint(m_dimension);
        for (int i = 0; i < m_dimension; i++) {
            final DataCell cell = row.getCell(fixedDataMdsIndices[i]);
            if (!cell.isMissing()) {
                final Double d = ((DoubleValue) cell).getDoubleValue();
                p.setElementAt(i, d);
            }
        }
        m_fixedPoints.put(row.getKey(), p);
    }
}

Example 53 with DoubleValue

use of org.knime.core.data.DoubleValue in project knime-core by knime.

the class DecisionTreeNodeSplitContinuous method addCoveredPattern.

/**
 * Add patterns given as a row of values if they fall within a specific
 * node. This node simply forwards this request to the appropriate child.
 *
 * @param cell the cell to be used for the split at this level
 * @param row input pattern
 * @param spec the corresponding table spec
 * @param weight the weight of the row (between 0.0 and 1.0)
 * @throws Exception if something went wrong (unknown attribute for example)
 */
@Override
public void addCoveredPattern(final DataCell cell, final DataRow row, final DataTableSpec spec, final double weight) throws Exception {
    double value = ((DoubleValue) cell).getDoubleValue();
    if (value <= m_threshold) {
        super.getChildNodeAt(0).addCoveredPattern(row, spec, weight);
    } else {
        super.getChildNodeAt(1).addCoveredPattern(row, spec, weight);
    }
    Color col = spec.getRowColor(row).getColor();
    addColorToMap(col, weight);
}

Example 54 with DoubleValue

use of org.knime.core.data.DoubleValue in project knime-core by knime.

the class DecisionTreeNodeSplitContinuous method addCoveredColor.

/**
 * Add colors for a pattern given as a row of values. This is a leaf so we
 * will simply add the color to our list.
 *
 * @param cell the cell to be used for the split at this level
 * @param row input pattern
 * @param spec the corresponding table spec
 * @param weight the weight of the row  (between 0.0 and 1.0)
 * @throws Exception if something went wrong (unknown attribute for example)
 */
@Override
public void addCoveredColor(final DataCell cell, final DataRow row, final DataTableSpec spec, final double weight) throws Exception {
    double value = ((DoubleValue) cell).getDoubleValue();
    if (value <= m_threshold) {
        super.getChildNodeAt(0).addCoveredColor(row, spec, weight);
    } else {
        super.getChildNodeAt(1).addCoveredColor(row, spec, weight);
    }
    Color col = spec.getRowColor(row).getColor();
    addColorToMap(col, weight);
}

Example 55 with DoubleValue

use of org.knime.core.data.DoubleValue in project knime-core by knime.

the class RPropNodeModel method configure.

/**
 * returns null.
 *
 * {@inheritDoc}
 */
@Override
protected PortObjectSpec[] configure(final PortObjectSpec[] inSpecs) throws InvalidSettingsException {
    if (m_classcol.getStringValue() != null) {
        List<String> learningCols = new LinkedList<String>();
        List<String> targetCols = new LinkedList<String>();
        boolean classcolinspec = false;
        for (DataColumnSpec colspec : (DataTableSpec) inSpecs[INDATA]) {
            if (!(colspec.getName().toString().compareTo(m_classcol.getStringValue()) == 0)) {
                if (!colspec.getType().isCompatible(DoubleValue.class)) {
                    throw new InvalidSettingsException("Only double columns for input");
                } else {
                    learningCols.add(colspec.getName());
                    DataColumnDomain domain = colspec.getDomain();
                    if (domain.hasBounds()) {
                        double lower = ((DoubleValue) domain.getLowerBound()).getDoubleValue();
                        double upper = ((DoubleValue) domain.getUpperBound()).getDoubleValue();
                        if (lower < 0 || upper > 1) {
                            setWarningMessage("Input data not normalized." + " Please consider using the " + "Normalizer Node first.");
                        }
                    }
                }
            } else {
                targetCols.add(colspec.getName());
                classcolinspec = true;
                // TODO: Check what happens to other values than double
                if (colspec.getType().isCompatible(DoubleValue.class)) {
                    // check if the values are in range [0,1]
                    DataColumnDomain domain = colspec.getDomain();
                    if (domain.hasBounds()) {
                        double lower = ((DoubleValue) domain.getLowerBound()).getDoubleValue();
                        double upper = ((DoubleValue) domain.getUpperBound()).getDoubleValue();
                        if (lower < 0 || upper > 1) {
                            throw new InvalidSettingsException("Domain range for regression in column " + colspec.getName() + " not in range [0,1]");
                        }
                    }
                }
            }
        }
        if (!classcolinspec) {
            throw new InvalidSettingsException("Class column " + m_classcol.getStringValue() + " not found in DataTableSpec");
        }
        return new PortObjectSpec[] { createPMMLPortObjectSpec(m_pmmlInEnabled ? (PMMLPortObjectSpec) inSpecs[1] : null, (DataTableSpec) inSpecs[0], learningCols, targetCols) };
    } else {
        throw new InvalidSettingsException("Class column not set");
    }
}