hex.tree.gbm

Class GBM

java.lang.Object

water.Iced

hex.ModelBuilder<M,P,O>

hex.tree.SharedTree<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

hex.tree.gbm.GBM

All Implemented Interfaces:

CalibrationHelper.ModelBuilderWithCalibration<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>, java.io.Externalizable, java.io.Serializable, java.lang.Cloneable, water.Freezable

public class GBM
extends SharedTree<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

Gradient Boosted Trees Based on "Elements of Statistical Learning, Second Edition, page 387"

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	GBM.DiffMinusMedianDiff

Nested classes/interfaces inherited from class hex.tree.SharedTree

SharedTree.Driver, SharedTree.FrameMap, SharedTree.ScoreBuildOneTree, SharedTree.SharedTreeDebugParams

Nested classes/interfaces inherited from class hex.ModelBuilder

hex.ModelBuilder.BuilderVisibility, hex.ModelBuilder.FilterCols, hex.ModelBuilder.ModelTrainingCoordinator, hex.ModelBuilder.ValidationMessage

Field Summary

Fields inherited from class hex.tree.SharedTree

_ics, _improvPerVar, _initialPrediction, _lastScoredTree, _model, _mtry, _mtry_per_tree, _ncols, _ntrees, _rand, _trainPredsCache, _validPredsCache, _validWorkspace, MAX_NTREES

Fields inherited from class hex.ModelBuilder

_coordinator, _desc, _eventPublisher, _fold, _input_parms, _job, _messages, _nclass, _offset, _orig_projection_array, _origDomains, _origNames, _origTrain, _parms, _priorClassDist, _removedCols, _response, _result, _startUpOnceModelBuilder, _train, _treatment, _valid, _vresponse, _weights

Constructor Summary

Constructors

Constructor and Description
GBM(boolean startup_once)
GBM(GBMModel.GBMParameters parms)
GBM(GBMModel.GBMParameters parms, water.Key<GBMModel> key)

Method Summary

Modifier and Type	Method and Description
hex.ModelCategory[]	can_build()
void	init(boolean expensive) Initialize the ModelBuilder, validating all arguments and preparing the training frame.
hex.PojoWriter	makePojoWriter(hex.Model<?,?,?> genericModel, hex.genmodel.MojoModel mojoModel)
protected int	nModelsInParallel(int folds)
protected void	raiseReproducibilityWarning(java.lang.String datasetName, int chunks)
protected double	score1(water.fvec.Chunk[] chks, double weight, double offset, double[] fs, int row)
protected hex.tree.gbm.GBM.GBMDriver	trainModelImpl() Start the GBM training Job on an F/J thread.

Methods inherited from class hex.tree.SharedTree

addCustomInfo, buildLayer, canLearnFromNAs, checkEarlyStoppingReproducibility, checkMemoryFootPrint_impl, chk_nids, chk_offset, chk_oobt, chk_resp, chk_tree, chk_weight, chk_work, computePriorClassDistribution, createModelSummaryTable, createScoringHistoryTable, createScoringHistoryTable, cv_canBuildMainModelInParallel, cv_computeAndSetOptimalParameters, cv_initStoppingParameters, cv_updateOptimalParameters, data_row, doScoringAndSaveModel, getCalibrationFrame, getInitialValue, getModelBuilder, getProblemType, getToEigenVec, haveMojo, havePojo, idx_nids, idx_offset, idx_oobt, idx_resp, idx_treatment, idx_tree, idx_weight, idx_work, ignoreInvalidColumns, isSupervised, isUplift, makeDecided, makeScoreExtension, providesVarImp, scoreZeroTrees, setCalibrationFrame, setDebugParams, shouldReorder, validateRowSampleRate, validWorkspace, vec_nids, vec_offset, vec_oobt, vec_resp, vec_tree, vec_weight, vec_work

Methods inherited from class hex.ModelBuilder

algoName, algos, builderVisibility, checkCustomMetricForEarlyStopping, checkDistributions, checkMemoryFootPrint, checkResponseVariable, clearInitState, clearValidationErrors, computeCrossValidation, cv_buildModels, cv_mainModelScores, cv_makeAggregateModelMetrics, cv_scoreCVModels, defaultKey, desiredChunks, dest, error_count, error, get, getMessagesByFieldAndSeverity, getName, getSysProperty, hasFoldCol, hasOffsetCol, hasTreatmentCol, hasWeightCol, hide, ignoreBadColumns, ignoreConstColumns, ignoreStringColumns, ignoreUuidColumns, info, init_adaptFrameToTrain, init_getNClass, initWorkspace, isClassifier, isResponseOptional, isStopped, javaName, logMe, make, make, make, makeCVMetrics, makeCVModelBuilder, makeParameters, message, nclasses, nFoldCV, nFoldWork, nModelsInParallel, nModelsInParallel, numSpecialCols, paramName, rebalance, remainingTimeSecs, response, schemaDirectory, separateFeatureVecs, setMaxRuntimeSecsForMainModel, setTrain, setValid, smallDataSize, stop_requested, timeout, train, trainModel, trainModel, trainModelNested, trainModelNested, trainModelOnH2ONode, valid, validateBinaryResponse, validateStoppingMetric, validationErrors, validationWarnings, vresponse, warn

Methods inherited from class water.Iced

asBytes, clone, copyOver, frozenType, read, readExternal, readJSON, reloadFromBytes, toJsonBytes, toJsonString, write, writeExternal, writeJSON

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

GBM

public GBM(GBMModel.GBMParameters parms)

GBM

public GBM(GBMModel.GBMParameters parms,
           water.Key<GBMModel> key)

GBM

public GBM(boolean startup_once)

Method Detail

can_build

public hex.ModelCategory[] can_build()

Specified by:

can_build in class hex.ModelBuilder<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

nModelsInParallel

protected int nModelsInParallel(int folds)

Overrides:

nModelsInParallel in class hex.ModelBuilder<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

trainModelImpl

protected hex.tree.gbm.GBM.GBMDriver trainModelImpl()

Start the GBM training Job on an F/J thread.

Specified by:

trainModelImpl in class hex.ModelBuilder<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

init

public void init(boolean expensive)

Initialize the ModelBuilder, validating all arguments and preparing the training frame. This call is expected to be overridden in the subclasses and each subclass will start with "super.init();". This call is made by the front-end whenever the GUI is clicked, and needs to be fast; heavy-weight prep needs to wait for the trainModel() call. Validate the learning rate and distribution family.

Overrides:

init in class SharedTree<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

score1

protected double score1(water.fvec.Chunk[] chks,
                        double weight,
                        double offset,
                        double[] fs,
                        int row)

Specified by:

score1 in class SharedTree<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

makePojoWriter

public hex.PojoWriter makePojoWriter(hex.Model<?,?,?> genericModel,
                                     hex.genmodel.MojoModel mojoModel)

Overrides:

makePojoWriter in class hex.ModelBuilder<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>

raiseReproducibilityWarning

protected void raiseReproducibilityWarning(java.lang.String datasetName,
                                           int chunks)

Overrides:

raiseReproducibilityWarning in class hex.ModelBuilder<GBMModel,GBMModel.GBMParameters,GBMModel.GBMOutput>