com.numericalmethod.suanshu.stats.hmm.mixture.distribution

Class BetaDistribution

java.lang.Object

com.numericalmethod.suanshu.stats.hmm.mixture.distribution.BetaDistribution

All Implemented Interfaces:

HMMDistribution, java.io.Serializable

public class BetaDistribution
extends java.lang.Object
implements HMMDistribution

The HMM states use the Beta distribution to model the observations.

See Also:

BetaDistribution, Wikipedia: Beta distribution, Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	BetaDistribution.Lambda the Beta distribution parameters

Constructor Summary

Constructors

Constructor and Description
BetaDistribution(BetaDistribution.Lambda[] lambda, boolean isAlphaEstimated, boolean isBetaEstimated, double epsilon, int maxIterations) Construct a Beta distribution for each state in the HMM model.
BetaDistribution(BetaDistribution.Lambda[] lambda, int maxIterations) Construct a Beta distribution for each state in the HMM model.

Method Summary

Modifier and Type	Method and Description
ProbabilityDistribution[]	getDistributions() Get the distributions (possibly differently parameterized) for all states.
BetaDistribution.Lambda[]	getMStepParams(double[] observations, Matrix u, java.lang.Object[] param0) Maximize, for each state, the log-likelihood of the distribution with respect to the observations and current estimators.
BetaDistribution.Lambda[]	getParams() Get the parameters, for each state, of the distribution.
RandomNumberGenerator[]	getRandomNumberGenerators() Get the random number generators corresponding to the distributions (possibly differently parameterized) for all states.
HMMDistribution	newEMDistribution(java.lang.Object[] param) Construct a new distribution from a set of parameters, one set per state.

Methods inherited from class java.lang.Object

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

Constructor Detail

BetaDistribution

public BetaDistribution(BetaDistribution.Lambda[] lambda,
                        boolean isAlphaEstimated,
                        boolean isBetaEstimated,
                        double epsilon,
                        int maxIterations)

Construct a Beta distribution for each state in the HMM model.

Parameters:

lambda - the distribution parameters

isAlphaEstimated - indicate whether parameter alpha is to be estimated

isBetaEstimated - indicate whether parameter beta is to be estimated

epsilon - a precision parameter: when a number |x| ≤ ε, it is considered 0

maxIterations - the maximum number of iterations in each M-step

BetaDistribution

public BetaDistribution(BetaDistribution.Lambda[] lambda,
                        int maxIterations)

Construct a Beta distribution for each state in the HMM model.

Parameters:

lambda - the distribution parameters

maxIterations - the maximum number of iterations in each M-step

Method Detail

getParams

public BetaDistribution.Lambda[] getParams()

Description copied from interface: HMMDistribution

Get the parameters, for each state, of the distribution.

Specified by:

getParams in interface HMMDistribution

Returns:

the parameters, for each state, of the distribution

getRandomNumberGenerators

public RandomNumberGenerator[] getRandomNumberGenerators()

Description copied from interface: HMMDistribution

Get the random number generators corresponding to the distributions (possibly differently parameterized) for all states.

Specified by:

getRandomNumberGenerators in interface HMMDistribution

Returns:

the random number generators corresponding to the distributions

getDistributions

public ProbabilityDistribution[] getDistributions()

Description copied from interface: HMMDistribution

Get the distributions (possibly differently parameterized) for all states.

Specified by:

getDistributions in interface HMMDistribution

Returns:

the distributions

getMStepParams

public BetaDistribution.Lambda[] getMStepParams(double[] observations,
                                                Matrix u,
                                                java.lang.Object[] param0)

Description copied from interface: HMMDistribution

Maximize, for each state, the log-likelihood of the distribution with respect to the observations and current estimators.

Specified by:

getMStepParams in interface HMMDistribution

Parameters:

observations - the observations

u - this is a log-transformed version of eq. (4.13) of Zucchini and MacDonald (2009), p. 65

param0 - the current estimators of the distribution parameters

Returns:

the MLE estimators

newEMDistribution

public HMMDistribution newEMDistribution(java.lang.Object[] param)

Description copied from interface: HMMDistribution

Construct a new distribution from a set of parameters, one set per state.

Specified by:

newEMDistribution in interface HMMDistribution

Parameters:

param - the parameters of the distribution, one set per state

Returns:

a new HMM distribution