Examples with BigGroupByTable org.knime.base.node.preproc.groupby.BigGroupByTable used on opensource projects

Example 1 with BigGroupByTable

use of org.knime.base.node.preproc.groupby.BigGroupByTable in project knime-core by knime.

the class CrosstabNodeModel method createGroupByTable.

/**
 * Create group-by table.
 * @param exec execution context
 * @param table input table to group
 * @param groupByCols column selected for group-by operation
 * @return table with group and aggregation columns
 * @throws CanceledExecutionException if the group-by table generation was
 *         canceled externally
 */
private final GroupByTable createGroupByTable(final ExecutionContext exec, final BufferedDataTable table, final List<String> groupByCols) throws CanceledExecutionException {
    final int maxUniqueVals = Integer.MAX_VALUE;
    final boolean enableHilite = m_settings.getEnableHiliting();
    final boolean retainOrder = false;
    final ColumnNamePolicy colNamePolicy = ColumnNamePolicy.AGGREGATION_METHOD_COLUMN_NAME;
    final GlobalSettings globalSettings = GlobalSettings.builder().setFileStoreFactory(FileStoreFactory.createWorkflowFileStoreFactory(exec)).setGroupColNames(groupByCols).setMaxUniqueValues(maxUniqueVals).setValueDelimiter(GlobalSettings.STANDARD_DELIMITER).setDataTableSpec(table.getDataTableSpec()).setNoOfRows(table.size()).setAggregationContext(AggregationContext.ROW_AGGREGATION).build();
    ColumnAggregator collAggregator = null;
    if (null != m_settings.getWeightColumn()) {
        final String weightColumn = m_settings.getWeightColumn();
        // the column aggregator for the weighting column
        final boolean inclMissing = false;
        final DataColumnSpec originalColSpec = table.getDataTableSpec().getColumnSpec(weightColumn);
        final OperatorColumnSettings opColSettings = new OperatorColumnSettings(inclMissing, originalColSpec);
        collAggregator = new ColumnAggregator(originalColSpec, new NonNegativeSumOperator(globalSettings, opColSettings), inclMissing);
    } else {
        // use any column, does not matter as long as it exists and
        // include missing is true;
        final boolean inclMissing = true;
        final DataColumnSpec originalColSpec = table.getDataTableSpec().getColumnSpec(groupByCols.get(0));
        final OperatorColumnSettings opColSettings = new OperatorColumnSettings(inclMissing, originalColSpec);
        collAggregator = new ColumnAggregator(originalColSpec, new CountOperator(globalSettings, opColSettings), inclMissing);
    }
    final GroupByTable resultTable = new BigGroupByTable(exec, table, groupByCols, new ColumnAggregator[] { collAggregator }, globalSettings, enableHilite, colNamePolicy, retainOrder);
    if (enableHilite) {
        setHiliteMapping(new DefaultHiLiteMapper(resultTable.getHiliteMapping()));
    }
    // check for skipped columns
    final String warningMsg = resultTable.getSkippedGroupsMessage(3, 3);
    if (warningMsg != null) {
        setWarningMessage(warningMsg);
    }
    return resultTable;
}

Example 2 with BigGroupByTable

use of org.knime.base.node.preproc.groupby.BigGroupByTable in project knime-core by knime.

the class NumericOutliersIntervalsCalculator method getGroupByTable.

/**
 * Constructs the group by table in accordance with the given settings.
 *
 * @param inTable the input data table
 * @param exec the execution context
 * @return the group by table w.r.t. the selected settings
 * @throws CanceledExecutionException if the user has canceled the execution
 */
private GroupByTable getGroupByTable(final BufferedDataTable inTable, final ExecutionContext exec) throws CanceledExecutionException {
    // get the global settings
    final GlobalSettings gSettings = getGlobalSettings(inTable);
    // create the column aggregators
    final ColumnAggregator[] agg = getAggretators(inTable.getDataTableSpec(), gSettings);
    // init and return the GroupByTable obeying the chosen memory settings
    final GroupByTable t;
    if (m_inMemory) {
        t = new MemoryGroupByTable(exec, inTable, Arrays.stream(m_groupColNames).collect(Collectors.toList()), agg, gSettings, false, COLUMN_NAME_POLICY, false);
    } else {
        t = new BigGroupByTable(exec, inTable, Arrays.stream(m_groupColNames).collect(Collectors.toList()), agg, gSettings, false, COLUMN_NAME_POLICY, false);
    }
    return t;
}