water.bindings.pojos

Class XGBoostParametersV3

java.lang.Object

water.bindings.pojos.SchemaV3

water.bindings.pojos.ModelParametersSchemaV3

water.bindings.pojos.XGBoostParametersV3

public class XGBoostParametersV3
extends ModelParametersSchemaV3

Field Summary

Fields

Modifier and Type	Field and Description
TreexgboostXGBoostModelXGBoostParametersBackend	backend Backend.
TreexgboostXGBoostModelXGBoostParametersBooster	booster Booster type
double	colsampleBylevel (same as col_sample_rate) Column sample rate (from 0.0 to 1.0)
double	colsampleBytree (same as col_sample_rate_per_tree) Column sample rate per tree (from 0.0 to 1.0)
double	colSampleRate (same as colsample_bylevel) Column sample rate (from 0.0 to 1.0)
double	colSampleRatePerTree (same as colsample_bytree) Column sample rate per tree (from 0.0 to 1.0)
TreexgboostXGBoostModelXGBoostParametersDMatrixType	dmatrixType Type of DMatrix.
double	eta (same as learn_rate) Learning rate (from 0.0 to 1.0)
float	gamma (same as min_split_improvement) Minimum relative improvement in squared error reduction for a split to happen
int	gpuId Which GPU to use.
TreexgboostXGBoostModelXGBoostParametersGrowPolicy	growPolicy Grow policy - depthwise is standard GBM, lossguide is LightGBM
double	learnRate (same as eta) Learning rate (from 0.0 to 1.0)
float	maxAbsLeafnodePred (same as max_delta_step) Maximum absolute value of a leaf node prediction
int	maxBins For tree_method=hist only: maximum number of bins
float	maxDeltaStep (same as max_abs_leafnode_pred) Maximum absolute value of a leaf node prediction
int	maxDepth Maximum tree depth.
int	maxLeaves For tree_method=hist only: maximum number of leaves
double	minChildWeight (same as min_rows) Fewest allowed (weighted) observations in a leaf.
float	minDataInLeaf For tree_method=hist only: the mininum data in a leaf to keep splitting
double	minRows (same as min_child_weight) Fewest allowed (weighted) observations in a leaf.
float	minSplitImprovement (same as gamma) Minimum relative improvement in squared error reduction for a split to happen
float	minSumHessianInLeaf For tree_method=hist only: the mininum sum of hessian in a leaf to keep splitting
int	nEstimators (same as ntrees) Number of trees.
TreexgboostXGBoostModelXGBoostParametersDartNormalizeType	normalizeType For booster=dart only: normalize_type
int	ntrees (same as n_estimators) Number of trees.
boolean	oneDrop For booster=dart only: one_drop
boolean	quietMode Enable quiet mode
float	rateDrop For booster=dart only: rate_drop (0..1)
float	regAlpha L1 regularization
float	regLambda L2 regularization
double	sampleRate (same as subsample) Row sample rate per tree (from 0.0 to 1.0)
TreexgboostXGBoostModelXGBoostParametersDartSampleType	sampleType For booster=dart only: sample_type
int	scoreTreeInterval Score the model after every so many trees.
long	seed Seed for pseudo random number generator (if applicable)
float	skipDrop For booster=dart only: skip_drop (0..1)
double	subsample (same as sample_rate) Row sample rate per tree (from 0.0 to 1.0)
TreexgboostXGBoostModelXGBoostParametersTreeMethod	treeMethod Tree method

Fields inherited from class water.bindings.pojos.ModelParametersSchemaV3

categoricalEncoding, checkpoint, distribution, foldAssignment, foldColumn, huberAlpha, ignoreConstCols, ignoredColumns, keepCrossValidationFoldAssignment, keepCrossValidationPredictions, maxCategoricalLevels, maxRuntimeSecs, modelId, nfolds, offsetColumn, parallelizeCrossValidation, quantileAlpha, responseColumn, scoreEachIteration, stoppingMetric, stoppingRounds, stoppingTolerance, trainingFrame, tweediePower, validationFrame, weightsColumn

Constructor Summary

Constructors

Constructor and Description
XGBoostParametersV3() Public constructor

Method Summary

Methods

Modifier and Type	Method and Description
java.lang.String	toString() Return the contents of this object as a JSON String.

Methods inherited from class java.lang.Object

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

Field Detail

ntrees

public int ntrees

(same as n_estimators) Number of trees.

nEstimators

public int nEstimators

(same as ntrees) Number of trees.

maxDepth

public int maxDepth

Maximum tree depth.

minRows

public double minRows

(same as min_child_weight) Fewest allowed (weighted) observations in a leaf.

minChildWeight

public double minChildWeight

(same as min_rows) Fewest allowed (weighted) observations in a leaf.

learnRate

public double learnRate

(same as eta) Learning rate (from 0.0 to 1.0)

eta

public double eta

(same as learn_rate) Learning rate (from 0.0 to 1.0)

sampleRate

public double sampleRate

(same as subsample) Row sample rate per tree (from 0.0 to 1.0)

subsample

public double subsample

(same as sample_rate) Row sample rate per tree (from 0.0 to 1.0)

colSampleRate

public double colSampleRate

(same as colsample_bylevel) Column sample rate (from 0.0 to 1.0)

colsampleBylevel

public double colsampleBylevel

(same as col_sample_rate) Column sample rate (from 0.0 to 1.0)

colSampleRatePerTree

public double colSampleRatePerTree

(same as colsample_bytree) Column sample rate per tree (from 0.0 to 1.0)

colsampleBytree

public double colsampleBytree

(same as col_sample_rate_per_tree) Column sample rate per tree (from 0.0 to 1.0)

maxAbsLeafnodePred

public float maxAbsLeafnodePred

(same as max_delta_step) Maximum absolute value of a leaf node prediction

maxDeltaStep

public float maxDeltaStep

(same as max_abs_leafnode_pred) Maximum absolute value of a leaf node prediction

scoreTreeInterval

public int scoreTreeInterval

Score the model after every so many trees. Disabled if set to 0.

seed

public long seed

Seed for pseudo random number generator (if applicable)

minSplitImprovement

public float minSplitImprovement

(same as gamma) Minimum relative improvement in squared error reduction for a split to happen

gamma

public float gamma

(same as min_split_improvement) Minimum relative improvement in squared error reduction for a split to happen

maxBins

public int maxBins

For tree_method=hist only: maximum number of bins

maxLeaves

public int maxLeaves

For tree_method=hist only: maximum number of leaves

minSumHessianInLeaf

public float minSumHessianInLeaf

For tree_method=hist only: the mininum sum of hessian in a leaf to keep splitting

minDataInLeaf

public float minDataInLeaf

For tree_method=hist only: the mininum data in a leaf to keep splitting

treeMethod

public TreexgboostXGBoostModelXGBoostParametersTreeMethod treeMethod

Tree method

growPolicy

public TreexgboostXGBoostModelXGBoostParametersGrowPolicy growPolicy

Grow policy - depthwise is standard GBM, lossguide is LightGBM

booster

public TreexgboostXGBoostModelXGBoostParametersBooster booster

Booster type

regLambda

public float regLambda

L2 regularization

regAlpha

public float regAlpha

L1 regularization

quietMode

public boolean quietMode

Enable quiet mode

sampleType

public TreexgboostXGBoostModelXGBoostParametersDartSampleType sampleType

For booster=dart only: sample_type

normalizeType

public TreexgboostXGBoostModelXGBoostParametersDartNormalizeType normalizeType

For booster=dart only: normalize_type

rateDrop

public float rateDrop

For booster=dart only: rate_drop (0..1)

oneDrop

public boolean oneDrop

For booster=dart only: one_drop

skipDrop

public float skipDrop

For booster=dart only: skip_drop (0..1)

dmatrixType

public TreexgboostXGBoostModelXGBoostParametersDMatrixType dmatrixType

Type of DMatrix. For sparse, NAs and 0 are treated equally.

backend

public TreexgboostXGBoostModelXGBoostParametersBackend backend

Backend. By default (auto), a GPU is used if available.

gpuId

public int gpuId

Which GPU to use.

Constructor Detail

XGBoostParametersV3

public XGBoostParametersV3()

Public constructor

Method Detail

toString

public java.lang.String toString()

Return the contents of this object as a JSON String.

Overrides:

toString in class ModelParametersSchemaV3