use of org.knime.base.node.mine.decisiontree2.model.DecisionTree in project knime-core by knime.
the class DecTreeLearnerGraphView method modelChanged.
/**
* {@inheritDoc}
*/
@Override
protected void modelChanged() {
DecisionTreeLearnerNodeModel model = this.getNodeModel();
if (model != null) {
m_hiLiteHdl.removeHiLiteListener(this);
DecisionTree dt = model.getDecisionTree();
if (dt != null) {
m_graph.setColorColumn(model.getDecisionTree().getColorColumn());
m_graph.setRootNode(dt.getRootNode());
// retrieve HiLiteHandler from Input port
m_hiLiteHdl = model.getInHiLiteHandler(DecisionTreeLearnerNodeModel.DATA_INPORT);
// and adjust menu entries for HiLite-ing
m_hiLiteMenu.setEnabled(m_hiLiteHdl != null);
m_hiLiteHdl.addHiLiteListener(this);
recreateHiLite();
} else {
m_graph.setColorColumn(null);
m_graph.setRootNode(null);
}
}
}
use of org.knime.base.node.mine.decisiontree2.model.DecisionTree in project knime-core by knime.
the class DecisionTreeLearnerNodeModel method execute.
/**
* Start of decision tree induction.
*
* @param exec the execution context for this run
* @param data the input data to build the decision tree from
* @return an empty data table array, as just a model is provided
* @throws Exception any type of exception, e.g. for cancellation,
* invalid input,...
* @see NodeModel#execute(BufferedDataTable[],ExecutionContext)
*/
@Override
protected PortObject[] execute(final PortObject[] data, final ExecutionContext exec) throws Exception {
// holds the warning message displayed after execution
StringBuilder warningMessageSb = new StringBuilder();
ParallelProcessing parallelProcessing = new ParallelProcessing(m_parallelProcessing.getIntValue());
if (LOGGER.isDebugEnabled()) {
LOGGER.debug("Number available threads: " + parallelProcessing.getMaxNumberThreads() + " used threads: " + parallelProcessing.getCurrentThreadsInUse());
}
exec.setProgress("Preparing...");
// check input data
assert (data != null && data[DATA_INPORT] != null);
BufferedDataTable inData = (BufferedDataTable) data[DATA_INPORT];
// get column with color information
String colorColumn = null;
for (DataColumnSpec s : inData.getDataTableSpec()) {
if (s.getColorHandler() != null) {
colorColumn = s.getName();
break;
}
}
// the data table must have more than 2 records
if (inData.getRowCount() <= 1) {
throw new IllegalArgumentException("Input data table must have at least 2 records!");
}
// get class column index
int classColumnIndex = inData.getDataTableSpec().findColumnIndex(m_classifyColumn.getStringValue());
assert classColumnIndex > -1;
// create initial In-Memory table
exec.setProgress("Create initial In-Memory table...");
InMemoryTableCreator tableCreator = new InMemoryTableCreator(inData, classColumnIndex, m_minNumberRecordsPerNode.getIntValue(), m_skipColumns.getBooleanValue());
InMemoryTable initialTable = tableCreator.createInMemoryTable(exec.createSubExecutionContext(0.05));
int removedRows = tableCreator.getRemovedRowsDueToMissingClassValue();
if (removedRows == inData.getRowCount()) {
throw new IllegalArgumentException("Class column contains only " + "missing values");
}
if (removedRows > 0) {
warningMessageSb.append(removedRows);
warningMessageSb.append(" rows removed due to missing class value;");
}
// the all over row count is used to report progress
m_alloverRowCount = initialTable.getSumOfWeights();
// set the finishing counter
// this counter will always be incremented when a leaf node is
// created, as this determines the recursion end and can thus
// be used for progress indication
m_finishedCounter = new AtomicDouble(0);
// get the number of attributes
m_numberAttributes = initialTable.getNumAttributes();
// create the quality measure
final SplitQualityMeasure splitQualityMeasure;
if (m_splitQualityMeasureType.getStringValue().equals(SPLIT_QUALITY_GINI)) {
splitQualityMeasure = new SplitQualityGini();
} else {
splitQualityMeasure = new SplitQualityGainRatio();
}
// build the tree
// before this set the node counter to 0
m_counter.set(0);
exec.setMessage("Building tree...");
DecisionTreeNode root = null;
root = buildTree(initialTable, exec, 0, splitQualityMeasure, parallelProcessing);
boolean isBinaryNominal = m_binaryNominalSplitMode.getBooleanValue();
boolean isFilterInvalidAttributeValues = m_filterNominalValuesFromParent.getBooleanValue();
if (isBinaryNominal && isFilterInvalidAttributeValues) {
// traverse tree nodes and remove from the children the attribute
// values that were filtered out further up in the tree. "Bug" 3124
root.filterIllegalAttributes(Collections.EMPTY_MAP);
}
// the decision tree model saved as PMML at the second out-port
DecisionTree decisionTree = new DecisionTree(root, m_classifyColumn.getStringValue(), /* strategy has to be set explicitly as the default in PMML is
none, which means rows with missing values are not
classified. */
PMMLMissingValueStrategy.LAST_PREDICTION);
decisionTree.setColorColumn(colorColumn);
// prune the tree
exec.setMessage("Prune tree with " + m_pruningMethod.getStringValue() + "...");
pruneTree(decisionTree);
// add highlight patterns and color information
exec.setMessage("Adding hilite and color info to tree...");
addHiliteAndColorInfo(inData, decisionTree);
LOGGER.info("Decision tree consisting of " + decisionTree.getNumberNodes() + " nodes created with pruning method " + m_pruningMethod.getStringValue());
// set the warning message if available
if (warningMessageSb.length() > 0) {
setWarningMessage(warningMessageSb.toString());
}
// reset the number available threads
parallelProcessing.reset();
parallelProcessing = null;
// no data out table is created -> return an empty table array
exec.setMessage("Creating PMML decision tree model...");
// handle the optional PMML input
PMMLPortObject inPMMLPort = (PMMLPortObject) data[1];
DataTableSpec inSpec = inData.getSpec();
PMMLPortObjectSpec outPortSpec = createPMMLPortObjectSpec(inPMMLPort == null ? null : inPMMLPort.getSpec(), inSpec);
PMMLPortObject outPMMLPort = new PMMLPortObject(outPortSpec, inPMMLPort, inData.getSpec());
outPMMLPort.addModelTranslater(new PMMLDecisionTreeTranslator(decisionTree));
m_decisionTree = decisionTree;
return new PortObject[] { outPMMLPort };
}
use of org.knime.base.node.mine.decisiontree2.model.DecisionTree in project knime-core by knime.
the class DecisionTreeLearnerNodeModel method loadInternals.
/**
* {@inheritDoc}
*/
@Override
protected void loadInternals(final File nodeInternDir, final ExecutionMonitor exec) throws IOException, CanceledExecutionException {
File internalsFile = new File(nodeInternDir, SAVE_INTERNALS_FILE_NAME);
if (!internalsFile.exists()) {
// file to load internals from not available
return;
}
BufferedInputStream in = new BufferedInputStream(new GZIPInputStream(new FileInputStream(internalsFile)));
ModelContentRO decisionTreeModel = ModelContent.loadFromXML(in);
try {
m_decisionTree = new DecisionTree(decisionTreeModel);
} catch (Exception e) {
// continue, but inform the user via a message
setWarningMessage("Internal model could not be loaded: " + e.getMessage() + ". The view will not display properly.");
}
}
use of org.knime.base.node.mine.decisiontree2.model.DecisionTree in project knime-core by knime.
the class Pruner method mdlPruning.
/**
* Prunes a {@link DecisionTree} according to the minimum description lenght
* (MDL) principle.
*
* @param decTree the decision tree to prune
*/
public static void mdlPruning(final DecisionTree decTree) {
// traverse the tree depth first (in-fix)
DecisionTreeNode root = decTree.getRootNode();
mdlPruningRecurse(root);
}
use of org.knime.base.node.mine.decisiontree2.model.DecisionTree in project knime-core by knime.
the class RegressionTreeModel method createDecisionTree.
public DecisionTree createDecisionTree(final DataTable sampleForHiliting) {
final DecisionTree result;
TreeModelRegression treeModel = getTreeModelRegression();
result = treeModel.createDecisionTree(m_metaData);
if (sampleForHiliting != null) {
final DataTableSpec dataSpec = sampleForHiliting.getDataTableSpec();
final DataTableSpec spec = getLearnAttributeSpec(dataSpec);
for (DataRow r : sampleForHiliting) {
try {
DataRow fullAttributeRow = createLearnAttributeRow(r, spec);
result.addCoveredPattern(fullAttributeRow, spec);
} catch (Exception e) {
// dunno what to do with that
NodeLogger.getLogger(getClass()).error("Error updating hilite info in tree view", e);
break;
}
}
}
return result;
}
Aggregations