Package org.tribuo.regression.xgboost

Class XGBoostOptions

java.lang.Object
org.tribuo.regression.xgboost.XGBoostOptions
All Implemented Interfaces:
com.oracle.labs.mlrg.olcut.config.Options
public class XGBoostOptions extends Object implements com.oracle.labs.mlrg.olcut.config.Options
CLI options for configuring an XGBoost regression trainer.

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    float
    alpha
    L1 regularization term for weights (default 0).
    int
    depth
    Max tree depth (default 6, range (0,inf]).
    int
    ensembleSize
    Number of trees in the ensemble.
    float
    eta
    Step size shrinkage parameter (default 0.3, range [0,1]).
    float
    gamma
    Minimum loss reduction to make a split (default 0, range [0,inf]).
    float
    lambda
    L2 regularization term for weights (default 1).
    float
    minWeight
    Minimum sum of instance weights needed in a leaf (default 1, range [0,inf]).
    int
    numThreads
    Number of threads to use (default 4, range (1, num hw threads)).
    boolean
    quiet
    Make the XGBoost training procedure quiet.
    XGBoostRegressionTrainer.RegressionType
    rType
    Regression type to use.
    float
    subsample
    Subsample size for each tree (default 1, range (0,1]).
    float
    subsampleFeatures
    Subsample features for each tree (default 1, range (0,1]).

    Fields inherited from interface com.oracle.labs.mlrg.olcut.config.Options

    header

  • Constructor Summary

    Constructors
    Constructor
    Description
    XGBoostOptions()
     

  • Method Summary

    Modifier and Type
    Method
    Description
    XGBoostRegressionTrainer
    getTrainer()
    Gets the configured XGBoostRegressionTrainer.

    Methods inherited from class java.lang.Object

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

    Methods inherited from interface com.oracle.labs.mlrg.olcut.config.Options

    getOptionsDescription

  • Field Details

    • rType

      @Option(longName="xgb-regression-metric", usage="Regression type to use. Defaults to LINEAR.") public XGBoostRegressionTrainer.RegressionType rType
      Regression type to use. Defaults to LINEAR.

    • ensembleSize

      @Option(longName="xgb-ensemble-size", usage="Number of trees in the ensemble.") public int ensembleSize
      Number of trees in the ensemble.

    • alpha

      @Option(longName="xgb-alpha", usage="L1 regularization term for weights (default 0).") public float alpha
      L1 regularization term for weights (default 0).

    • minWeight

      @Option(longName="xgb-min-weight", usage="Minimum sum of instance weights needed in a leaf (default 1, range [0,inf]).") public float minWeight
      Minimum sum of instance weights needed in a leaf (default 1, range [0,inf]).

    • depth

      @Option(longName="xgb-max-depth", usage="Max tree depth (default 6, range (0,inf]).") public int depth
      Max tree depth (default 6, range (0,inf]).

    • eta

      @Option(longName="xgb-eta", usage="Step size shrinkage parameter (default 0.3, range [0,1]).") public float eta
      Step size shrinkage parameter (default 0.3, range [0,1]).

    • subsampleFeatures

      @Option(longName="xgb-subsample-features", usage="Subsample features for each tree (default 1, range (0,1]).") public float subsampleFeatures
      Subsample features for each tree (default 1, range (0,1]).

    • gamma

      @Option(longName="xgb-gamma", usage="Minimum loss reduction to make a split (default 0, range [0,inf]).") public float gamma
      Minimum loss reduction to make a split (default 0, range [0,inf]).

    • lambda

      @Option(longName="xgb-lambda", usage="L2 regularization term for weights (default 1).") public float lambda
      L2 regularization term for weights (default 1).

    • quiet

      @Option(longName="xgb-quiet", usage="Make the XGBoost training procedure quiet.") public boolean quiet
      Make the XGBoost training procedure quiet.

    • subsample

      @Option(longName="xgb-subsample", usage="Subsample size for each tree (default 1, range (0,1]).") public float subsample
      Subsample size for each tree (default 1, range (0,1]).

    • numThreads

      @Option(longName="xgb-num-threads", usage="Number of threads to use (default 4, range (1, num hw threads)).") public int numThreads
      Number of threads to use (default 4, range (1, num hw threads)).

  • Constructor Details

    • XGBoostOptions

      public XGBoostOptions()

  • Method Details

    • getTrainer

      public XGBoostRegressionTrainer getTrainer()
      Gets the configured XGBoostRegressionTrainer.
      Returns:
      The configured trainer.