weka.classifiers.mi.miti

Class AlgorithmConfiguration

java.lang.Object

weka.classifiers.mi.miti.AlgorithmConfiguration

public class AlgorithmConfiguration
extends java.lang.Object

Stores parameters that determine the configuration of the algorithm.

Field Summary

Fields

Modifier and Type	Field and Description
int	attributeSplitChoices The number of top ranked attribute splits to randomly pick from (default value 1 has no randomness)
int	attributesToSplit The number of attributes randomly selected to find the best from.
double	bagCountMultiplier The value used to determine the influence of instance counts when using bag counts.
int	kBEPPConstant The constant used to scale the influence of a node's size on its split score.
int	method The method used to score a split (1 = Gini, 2 = Max BEPP, 3 = SSBEPP)
boolean	unbiasedEstimate Determines whether an unbiased score is used to estimate the proportion of positives
boolean	useBagStatistics Determines whether bag stats are used to score splits, or instance stats.

Constructor Summary

Constructors

Constructor and Description
AlgorithmConfiguration() Default constructor that assigns default values.
AlgorithmConfiguration(int method, boolean unbiasedEstimate, int kBEPPConstant, boolean bagStatistics, double bagCountMultiplier, int attributesToSplit, int attributeSplitChoices) Constructor that sets algorithm parameters based on arguments.

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

method

public int method

The method used to score a split (1 = Gini, 2 = Max BEPP, 3 = SSBEPP)

unbiasedEstimate

public boolean unbiasedEstimate

Determines whether an unbiased score is used to estimate the proportion of positives

kBEPPConstant

public int kBEPPConstant

The constant used to scale the influence of a node's size on its split score.

useBagStatistics

public boolean useBagStatistics

Determines whether bag stats are used to score splits, or instance stats.

bagCountMultiplier

public double bagCountMultiplier

The value used to determine the influence of instance counts when using bag counts. Multiplier is used as M^(instance count for bag), where M is the bag count multiplier. Default value 0.5

attributesToSplit

public int attributesToSplit

The number of attributes randomly selected to find the best from. -1: All attributes, -2: Square root of total attribute count

attributeSplitChoices

public int attributeSplitChoices

The number of top ranked attribute splits to randomly pick from (default value 1 has no randomness)

Constructor Detail

AlgorithmConfiguration

public AlgorithmConfiguration()

Default constructor that assigns default values.

AlgorithmConfiguration

public AlgorithmConfiguration(int method,
                              boolean unbiasedEstimate,
                              int kBEPPConstant,
                              boolean bagStatistics,
                              double bagCountMultiplier,
                              int attributesToSplit,
                              int attributeSplitChoices)

Constructor that sets algorithm parameters based on arguments.