use of org.knime.core.data.container.DataContainer in project knime-core by knime.
the class ConditionalBoxPlotNodeModel method execute.
/**
* {@inheritDoc}
*/
@Override
protected BufferedDataTable[] execute(final BufferedDataTable[] inData, final ExecutionContext exec) throws Exception {
m_statistics = new LinkedHashMap<DataColumnSpec, double[]>();
m_mildOutliers = new LinkedHashMap<String, Map<Double, Set<RowKey>>>();
m_extremeOutliers = new LinkedHashMap<String, Map<Double, Set<RowKey>>>();
double nrRows = inData[0].size();
int rowCount = 0;
int numericIndex = inData[0].getDataTableSpec().findColumnIndex(m_settings.numericColumn());
int nominalIndex = inData[0].getDataTableSpec().findColumnIndex(m_settings.nominalColumn());
Map<String, Map<Double, Set<RowKey>>> data = new LinkedHashMap<String, Map<Double, Set<RowKey>>>();
// some default values .. if one column only has missing values.
for (DataCell d : inData[0].getDataTableSpec().getColumnSpec(nominalIndex).getDomain().getValues()) {
String name = ((StringValue) d).getStringValue();
m_mildOutliers.put(name, new HashMap<Double, Set<RowKey>>());
m_extremeOutliers.put(name, new HashMap<Double, Set<RowKey>>());
}
for (DataRow r : inData[0]) {
exec.checkCanceled();
exec.setProgress(rowCount++ / nrRows, "Separating...");
if (!m_settings.showMissingValues()) {
if (r.getCell(nominalIndex).isMissing()) {
// missing cell in nominal values is unwanted?
continue;
}
}
String nominal = replaceSpaces(r.getCell(nominalIndex).toString());
if (r.getCell(numericIndex).isMissing()) {
// ignore missing cells in numeric column
continue;
}
DoubleValue numeric = (DoubleValue) r.getCell(numericIndex);
Map<Double, Set<RowKey>> map = data.get(nominal);
if (map == null) {
map = new LinkedHashMap<Double, Set<RowKey>>();
}
Set<RowKey> set = map.get(numeric.getDoubleValue());
if (set == null) {
set = new HashSet<RowKey>();
}
set.add(r.getKey());
map.put(numeric.getDoubleValue(), set);
data.put(nominal, map);
}
List<String> keys = new ArrayList<String>(data.keySet());
boolean ignoreMissingValues = false;
if (m_settings.showMissingValues() && !keys.contains(DataType.getMissingCell().toString())) {
// we promised to create data for missing values..
// if there aren't any.. we have to create them ourselves
setWarningMessage("No missing values found.");
ignoreMissingValues = true;
}
Collections.sort(keys);
DataColumnSpec[] colSpecs = createColumnSpec(inData[0].getDataTableSpec().getColumnSpec(nominalIndex), ignoreMissingValues);
if (keys.size() == 0) {
setWarningMessage("All classes are empty.");
}
int dataSetNr = 0;
// for (String d : keys) {
for (DataColumnSpec dcs : colSpecs) {
String d = dcs.getName();
if (data.get(d) == null || keys.size() == 0) {
dataSetNr++;
continue;
}
exec.checkCanceled();
exec.setProgress(dataSetNr / (double) keys.size(), "Creating statistics");
Map<Double, Set<RowKey>> extremeOutliers = new LinkedHashMap<Double, Set<RowKey>>();
Map<Double, Set<RowKey>> mildOutliers = new LinkedHashMap<Double, Set<RowKey>>();
double[] stats = calculateStatistic(data.get(d), mildOutliers, extremeOutliers);
double minimum = stats[BoxPlotNodeModel.MIN];
double maximum = stats[BoxPlotNodeModel.MAX];
DataColumnSpecCreator creator = new DataColumnSpecCreator(colSpecs[dataSetNr]);
creator.setDomain(new DataColumnDomainCreator(new DoubleCell(minimum), new DoubleCell(maximum)).createDomain());
colSpecs[dataSetNr] = creator.createSpec();
m_statistics.put(colSpecs[dataSetNr], stats);
m_mildOutliers.put(d, mildOutliers);
m_extremeOutliers.put(d, extremeOutliers);
dataSetNr++;
}
DataTableSpec dts = new DataTableSpec("MyTempTable", colSpecs);
DataContainer cont = new DataContainer(dts);
cont.close();
m_dataArray = new DefaultDataArray(cont.getTable(), 1, 2);
cont.dispose();
if (ignoreMissingValues) {
DataColumnSpec[] temp = new DataColumnSpec[colSpecs.length + 1];
DataColumnSpec missing = new DataColumnSpecCreator(DataType.getMissingCell().toString(), DataType.getMissingCell().getType()).createSpec();
int i = 0;
while (missing.getName().compareTo(colSpecs[i].getName()) > 0) {
temp[i] = colSpecs[i];
i++;
}
temp[i++] = missing;
while (i < temp.length) {
temp[i] = colSpecs[i - 1];
i++;
}
colSpecs = temp;
}
/* Save inSpec of the numeric column to provide the view a way to
* consider the input domain for normalization. */
m_numColSpec = inData[0].getDataTableSpec().getColumnSpec(numericIndex);
return new BufferedDataTable[] { createOutputTable(inData[0].getDataTableSpec(), colSpecs, exec).getTable() };
}
use of org.knime.core.data.container.DataContainer in project knime-core by knime.
the class LiftCalculator method calculateLiftTables.
/**
* Calculates the tables necessary for displaying a lift chart.
* @param table the data table
* @param exec the execution context to report progress to
* @return warning messages or null
* @throws CanceledExecutionException when the user cancels the execution
*/
public String calculateLiftTables(final BufferedDataTable table, final ExecutionContext exec) throws CanceledExecutionException {
int predColIndex = table.getDataTableSpec().findColumnIndex(m_responseColumn);
String warning = null;
List<String> inclList = new LinkedList<String>();
inclList.add(m_probabilityColumn);
int probColInd = table.getDataTableSpec().findColumnIndex(m_probabilityColumn);
boolean[] order = new boolean[] { false };
m_sorted = new SortedTable(table, inclList, order, exec);
long totalResponses = 0;
double partWidth = m_intervalWidth;
int nrParts = (int) Math.ceil(100.0 / partWidth);
List<Integer> positiveResponses = new LinkedList<Integer>();
int rowIndex = 0;
for (DataRow row : m_sorted) {
if (row.getCell(predColIndex).isMissing() || row.getCell(probColInd).isMissing()) {
if (row.getCell(predColIndex).isMissing()) {
// miss. values in class column we always ignore
continue;
}
if (m_ignoreMissingValues) {
continue;
} else {
warning = "Table contains missing values.";
}
}
String response = ((StringValue) row.getCell(predColIndex)).getStringValue().trim();
if (response.equalsIgnoreCase(m_responseLabel)) {
totalResponses++;
positiveResponses.add(rowIndex);
}
rowIndex++;
}
int[] counter = new int[nrParts];
int partWidthAbsolute = (int) Math.ceil(rowIndex / (double) nrParts);
double avgResponse = (double) positiveResponses.size() / rowIndex;
for (int rIndex : positiveResponses) {
int index = rIndex / partWidthAbsolute;
counter[index]++;
}
DataColumnSpec[] colSpec = new DataColumnSpec[3];
colSpec[0] = new DataColumnSpecCreator("Lift", DoubleCell.TYPE).createSpec();
colSpec[1] = new DataColumnSpecCreator("Baseline", DoubleCell.TYPE).createSpec();
colSpec[2] = new DataColumnSpecCreator("Cumulative Lift", DoubleCell.TYPE).createSpec();
DataTableSpec tableSpec = new DataTableSpec(colSpec);
// new DataContainer(tableSpec);
DataContainer cont = exec.createDataContainer(tableSpec);
colSpec = new DataColumnSpec[2];
colSpec[0] = new DataColumnSpecCreator("Actual", DoubleCell.TYPE).createSpec();
colSpec[1] = new DataColumnSpecCreator("Baseline", DoubleCell.TYPE).createSpec();
tableSpec = new DataTableSpec(colSpec);
// new DataContainer(tableSpec);
DataContainer responseCont = exec.createDataContainer(tableSpec);
long cumulativeCounter = 0;
responseCont.addRowToTable(new DefaultRow(new RowKey("0"), 0.0, 0.0));
for (int i = 0; i < counter.length; i++) {
cumulativeCounter += counter[i];
double responseRate = (double) counter[i] / partWidthAbsolute;
double lift = responseRate / avgResponse;
double cumResponseRate = (double) cumulativeCounter / totalResponses;
long number = partWidthAbsolute * (i + 1);
// well.. rounding problems
if (number > rowIndex) {
number = rowIndex;
}
double cumulativeLift = // (double)cumulativeCounter / (partWidthAbsolute * (i + 1));
(double) cumulativeCounter / number;
cumulativeLift /= avgResponse;
// cumulativeLift = lifts / (i+1);
double rowKey = ((i + 1) * partWidth);
if (rowKey > 100) {
rowKey = 100;
}
cont.addRowToTable(new DefaultRow(new RowKey("" + rowKey), lift, 1.0, cumulativeLift));
double cumBaseline = (i + 1) * partWidth;
if (cumBaseline > 100) {
cumBaseline = 100;
}
responseCont.addRowToTable(new DefaultRow(new RowKey("" + rowKey), cumResponseRate * 100, cumBaseline));
}
cont.close();
responseCont.close();
m_lift = (BufferedDataTable) cont.getTable();
m_response = (BufferedDataTable) responseCont.getTable();
return warning;
}
use of org.knime.core.data.container.DataContainer in project knime-core by knime.
the class BoxplotCalculator method calculateMultiple.
/**
* Calculates the necessary statistics for a non-conditional boxplot.
* @param table the input data
* @param numCol array of names of numeric columns to plot
* @param exec Execution context to report progress to
* @return LinkedHashMap with the column name as key and statistics as value
* @throws CanceledExecutionException when the user cancels the execution
*/
public LinkedHashMap<String, BoxplotStatistics> calculateMultiple(final BufferedDataTable table, final String[] numCol, final ExecutionContext exec) throws CanceledExecutionException {
DataTableSpec spec = table.getSpec();
int[] numColIdxs = new int[numCol.length];
for (int i = 0; i < numCol.length; i++) {
numColIdxs[i] = spec.findColumnIndex(numCol[i]);
}
LinkedHashMap<String, DataContainer> containers = new LinkedHashMap<String, DataContainer>();
for (int i = 0; i < numCol.length; i++) {
containers.put(numCol[i], exec.createDataContainer(new DataTableSpec(new String[] { "col" }, new DataType[] { DoubleCell.TYPE })));
}
ExecutionContext subExec = exec.createSilentSubExecutionContext(0.7);
long[] numMissValPerCol = new long[numCol.length];
int count = 0;
for (DataRow row : table) {
exec.checkCanceled();
subExec.setProgress((double) count++ / table.size());
for (int i = 0; i < numCol.length; i++) {
DataCell cell = row.getCell(numColIdxs[i]);
if (!cell.isMissing()) {
containers.get(numCol[i]).addRowToTable(new DefaultRow(row.getKey(), cell));
} else {
numMissValPerCol[i]++;
}
}
}
LinkedHashMap<String, BoxplotStatistics> statsMap = new LinkedHashMap<>();
ExecutionContext subExec2 = exec.createSilentSubExecutionContext(1.0);
count = 0;
List<String> excludedDataColList = new ArrayList<String>();
for (Entry<String, DataContainer> entry : containers.entrySet()) {
exec.checkCanceled();
subExec2.setProgress((double) count++ / containers.size());
Set<Outlier> extremeOutliers = new HashSet<Outlier>();
Set<Outlier> mildOutliers = new HashSet<Outlier>();
entry.getValue().close();
BufferedDataTable catTable = (BufferedDataTable) entry.getValue().getTable();
if (catTable.size() == 0) {
excludedDataColList.add(entry.getKey());
continue;
}
SortedTable st = new SortedTable(catTable, new Comparator<DataRow>() {
@Override
public int compare(final DataRow o1, final DataRow o2) {
DataCell c1 = o1.getCell(0);
DataCell c2 = o2.getCell(0);
double d1 = ((DoubleValue) c1).getDoubleValue();
double d2 = ((DoubleValue) c2).getDoubleValue();
if (d1 == d2) {
return 0;
} else {
return d1 < d2 ? -1 : 1;
}
}
}, false, exec);
double min = 0, max = 0, q1 = 0, q3 = 0, median = 0;
boolean dq1 = catTable.size() % 4 == 0;
long q1Idx = catTable.size() / 4;
boolean dq3 = 3 * catTable.size() % 4 == 0;
long q3Idx = 3 * catTable.size() / 4;
boolean dMedian = catTable.size() % 2 == 0;
long medianIdx = catTable.size() / 2;
int counter = 0;
for (DataRow row : st) {
double val = ((DoubleValue) row.getCell(0)).getDoubleValue();
if (counter == 0) {
min = val;
}
if (counter == catTable.size() - 1) {
max = val;
}
if (counter == q1Idx - 1 && dq1) {
q1 = val;
}
if (counter == q1Idx || (counter == 0 && st.size() <= 3)) {
if (dq1) {
q1 = (q1 + val) / 2.0;
} else {
q1 = val;
}
}
if (counter == medianIdx - 1 && dMedian) {
median = val;
}
if (counter == medianIdx) {
if (dMedian) {
median = (median + val) / 2;
} else {
median = val;
}
}
if (counter == q3Idx - 1 && dq3) {
q3 = val;
}
if (counter == q3Idx || (counter == st.size() - 1 && st.size() <= 3)) {
if (dq3) {
q3 = (q3 + val) / 2.0;
} else {
q3 = val;
}
}
counter++;
}
double iqr = q3 - q1;
double lowerWhisker = min;
double upperWhisker = max;
double upperWhiskerFence = q3 + (1.5 * iqr);
double lowerWhiskerFence = q1 - (1.5 * iqr);
double lowerFence = q1 - (3 * iqr);
double upperFence = q3 + (3 * iqr);
for (DataRow row : st) {
double value = ((DoubleValue) row.getCell(0)).getDoubleValue();
String rowKey = row.getKey().getString();
if (value < lowerFence) {
extremeOutliers.add(new Outlier(value, rowKey));
} else if (value < lowerWhiskerFence) {
mildOutliers.add(new Outlier(value, rowKey));
} else if (lowerWhisker < lowerWhiskerFence && value >= lowerWhiskerFence) {
lowerWhisker = value;
} else if (value <= upperWhiskerFence) {
upperWhisker = value;
} else if (value > upperFence) {
extremeOutliers.add(new Outlier(value, rowKey));
} else if (value > upperWhiskerFence) {
mildOutliers.add(new Outlier(value, rowKey));
}
}
statsMap.put(entry.getKey(), new BoxplotStatistics(mildOutliers, extremeOutliers, min, max, lowerWhisker, q1, median, q3, upperWhisker));
}
// missing values part
m_excludedDataCols = excludedDataColList.toArray(new String[excludedDataColList.size()]);
m_numMissValPerCol = new LinkedHashMap<String, Long>();
for (int i = 0; i < numCol.length; i++) {
if (numMissValPerCol[i] > 0 && !excludedDataColList.contains(numCol[i])) {
m_numMissValPerCol.put(numCol[i], numMissValPerCol[i]);
}
}
return statsMap;
}
use of org.knime.core.data.container.DataContainer in project knime-core by knime.
the class MissingValueHandlerNodeModel method execute.
/**
* {@inheritDoc}
*/
@Override
protected PortObject[] execute(final PortObject[] inData, final ExecutionContext exec) throws Exception {
BufferedDataTable inTable = (BufferedDataTable) inData[0];
DataTableSpec inSpec = inTable.getDataTableSpec();
MissingCellReplacingDataTable mvTable = new MissingCellReplacingDataTable(inSpec, m_settings);
// Calculate the statistics
exec.setMessage("Calculating statistics");
mvTable.init(inTable, exec.createSubExecutionContext(0.5));
long rowCounter = 0;
final long numOfRows = inTable.size();
DataContainer container = exec.createDataContainer(mvTable.getDataTableSpec());
ExecutionContext tableSubExec = exec.createSubExecutionContext(0.4);
exec.setMessage("Replacing missing values");
for (DataRow row : mvTable) {
tableSubExec.checkCanceled();
if (row != null) {
tableSubExec.setProgress(++rowCounter / (double) numOfRows, "Processed row " + rowCounter + "/" + numOfRows + " (\"" + row.getKey() + "\")");
container.addRowToTable(row);
} else {
tableSubExec.setProgress(++rowCounter / (double) numOfRows, "Processed row " + rowCounter + "/" + numOfRows);
}
}
container.close();
// Collect warning messages
String warnings = mvTable.finish();
// Handle the warnings
if (warnings.length() > 0) {
setWarningMessage(warnings);
}
exec.setMessage("Generating PMML");
// Init PMML output port
PMMLPortObject pmmlPort = new PMMLPortObject(new PMMLPortObjectSpecCreator(inSpec).createSpec());
pmmlPort.addModelTranslater(mvTable.getPMMLTranslator());
return new PortObject[] { (BufferedDataTable) container.getTable(), pmmlPort };
}
use of org.knime.core.data.container.DataContainer in project knime-core by knime.
the class MappingTableInterpolationStatistic method init.
/**
* {@inheritDoc}
*/
@Override
protected void init(final DataTableSpec spec, final int amountOfColumns) {
m_index = spec.findColumnIndex(m_columnName);
m_nextCells = new DataContainer(new DataTableSpec(new DataColumnSpecCreator("value", spec.getColumnSpec(m_index).getType()).createSpec()));
m_previous = DataType.getMissingCell();
}
Aggregations