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
KeyValueV3[]	monotoneConstraints A mapping representing monotonic constraints.
int	nEstimators (same as ntrees) Number of trees.
TreexgboostXGBoostModelXGBoostParametersDartNormalizeType	normalizeType For booster=dart only: normalize_type
int	nthread Number of parallel threads that can be used to run XGBoost.
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, customMetricFunc, distribution, exportCheckpointsDir, foldAssignment, foldColumn, huberAlpha, ignoreConstCols, ignoredColumns, keepCrossValidationFoldAssignment, keepCrossValidationModels, 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

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

@SerializedName(value="n_estimators")
public int nEstimators

(same as ntrees) Number of trees.

maxDepth

@SerializedName(value="max_depth")
public int maxDepth

Maximum tree depth.

minRows

@SerializedName(value="min_rows")
public double minRows

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

minChildWeight

@SerializedName(value="min_child_weight")
public double minChildWeight

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

learnRate

@SerializedName(value="learn_rate")
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

@SerializedName(value="sample_rate")
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

@SerializedName(value="col_sample_rate")
public double colSampleRate

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

colsampleBylevel

@SerializedName(value="colsample_bylevel")
public double colsampleBylevel

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

colSampleRatePerTree

@SerializedName(value="col_sample_rate_per_tree")
public double colSampleRatePerTree

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

colsampleBytree

@SerializedName(value="colsample_bytree")
public double colsampleBytree

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

monotoneConstraints

@SerializedName(value="monotone_constraints")
public KeyValueV3[] monotoneConstraints

A mapping representing monotonic constraints. Use +1 to enforce an increasing constraint and -1 to specify a decreasing constraint.

maxAbsLeafnodePred

@SerializedName(value="max_abs_leafnode_pred")
public float maxAbsLeafnodePred

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

maxDeltaStep

@SerializedName(value="max_delta_step")
public float maxDeltaStep

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

scoreTreeInterval

@SerializedName(value="score_tree_interval")
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

@SerializedName(value="min_split_improvement")
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

nthread

public int nthread

Number of parallel threads that can be used to run XGBoost. Cannot exceed H2O cluster limits (-nthreads parameter). Defaults to maximum available

maxBins

@SerializedName(value="max_bins")
public int maxBins

For tree_method=hist only: maximum number of bins

maxLeaves

@SerializedName(value="max_leaves")
public int maxLeaves

For tree_method=hist only: maximum number of leaves

minSumHessianInLeaf

@SerializedName(value="min_sum_hessian_in_leaf")
public float minSumHessianInLeaf

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

minDataInLeaf

@SerializedName(value="min_data_in_leaf")
public float minDataInLeaf

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

treeMethod

@SerializedName(value="tree_method")
public TreexgboostXGBoostModelXGBoostParametersTreeMethod treeMethod

Tree method

growPolicy

@SerializedName(value="grow_policy")
public TreexgboostXGBoostModelXGBoostParametersGrowPolicy growPolicy

Grow policy - depthwise is standard GBM, lossguide is LightGBM

booster

public TreexgboostXGBoostModelXGBoostParametersBooster booster

Booster type

regLambda

@SerializedName(value="reg_lambda")
public float regLambda

L2 regularization

regAlpha

@SerializedName(value="reg_alpha")
public float regAlpha

L1 regularization

quietMode

@SerializedName(value="quiet_mode")
public boolean quietMode

Enable quiet mode

sampleType

@SerializedName(value="sample_type")
public TreexgboostXGBoostModelXGBoostParametersDartSampleType sampleType

For booster=dart only: sample_type

normalizeType

@SerializedName(value="normalize_type")
public TreexgboostXGBoostModelXGBoostParametersDartNormalizeType normalizeType

For booster=dart only: normalize_type

rateDrop

@SerializedName(value="rate_drop")
public float rateDrop

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

oneDrop

@SerializedName(value="one_drop")
public boolean oneDrop

For booster=dart only: one_drop

skipDrop

@SerializedName(value="skip_drop")
public float skipDrop

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

dmatrixType

@SerializedName(value="dmatrix_type")
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

@SerializedName(value="gpu_id")
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