org.encog.ml.hmm

Class HiddenMarkovModel

java.lang.Object

org.encog.ml.BasicML

org.encog.ml.hmm.HiddenMarkovModel

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, MLMethod, MLProperties, MLStateSequence

public class HiddenMarkovModel
extends BasicML
implements MLStateSequence, java.io.Serializable, java.lang.Cloneable

A Hidden Markov Model (HMM) is a Machine Learning Method that allows for predictions to be made about the hidden states and observations of a given system over time. A HMM can be thought of as a simple dynamic Bayesian network. The HMM is dynamic as it deals with changes that unfold over time. The Hidden Markov Model is made up of a number of states and observations. A simple example might be the state of the economy. There are three hidden states, such as bull market, bear market and level. We do not know which state we are currently in. However, there are observations that can be made such as interest rate and the level of the S&P500. The HMM learns what state we are in by seeing how the observations change over time. The HMM is only in one state at a given time. There is a percent probability that the HMM will move from one state to any of the other states. These probabilities are arranged in a grid, and are called the state transition probabilities. Observations can be discrete or continuous. These observations allow the HMM to predict state transitions. The HMM can handle single-value or multivariate observations. http://www.heatonresearch.com/wiki/Hidden_Markov_Model Rabiner, Juang, An introduction to Hidden Markov Models, IEEE ASSP Mag.,pp 4-16, June 1986. Baum, L. E.; Petrie, T. (1966). "Statistical Inference for Probabilistic Functions of Finite State Markov Chains" The Annals of Mathematical Statistics 37 (6): 1554-1563.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.String	TAG_COVARIANCE
static java.lang.String	TAG_DIST_TYPE
static java.lang.String	TAG_ITEMS
static java.lang.String	TAG_MEAN
static java.lang.String	TAG_PI
static java.lang.String	TAG_PROBABILITIES
static java.lang.String	TAG_STATES
static java.lang.String	TAG_TRANSITION

Constructor Summary

Constructors

Constructor and Description
HiddenMarkovModel(int states) Construct a discrete HMM with the specified number of states.
HiddenMarkovModel(int theStates, int theItems)
HiddenMarkovModel(int theStates, int[] theItems)

Method Summary

Modifier and Type	Method and Description
HiddenMarkovModel	clone()
HiddenMarkovModel	cloneStructure()
StateDistribution	createNewDistribution()
int[]	getItems()
double[]	getPi()
double	getPi(int i)
int	getStateCount()
StateDistribution	getStateDistribution(int i)
int[]	getStatesForSequence(MLDataSet seq) Get the sates for the given sequence.
double[][]	getTransitionProbability()
double	getTransitionProbability(int i, int j)
boolean	isContinuous()
boolean	isDiscrete()
double	lnProbability(MLDataSet seq)
double	probability(MLDataSet seq) Determine the probability of the specified sequence.
double	probability(MLDataSet seq, int[] states) Determine the probability for the specified sequence and states.
void	setPi(double[] data)
void	setPi(int i, double value)
void	setStateDistribution(int i, StateDistribution dist)
void	setTransitionProbability(double[][] data)
void	setTransitionProbability(int i, int j, double value)
void	updateProperties() Update any objeccts when a property changes.

Methods inherited from class org.encog.ml.BasicML

getProperties, getPropertyDouble, getPropertyLong, getPropertyString, setProperty, setProperty, setProperty

Methods inherited from class java.lang.Object

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

Field Detail

TAG_STATES

public static final java.lang.String TAG_STATES

See Also:

Constant Field Values

TAG_ITEMS

public static final java.lang.String TAG_ITEMS

See Also:

Constant Field Values

TAG_PI

public static final java.lang.String TAG_PI

See Also:

Constant Field Values

TAG_TRANSITION

public static final java.lang.String TAG_TRANSITION

See Also:

Constant Field Values

TAG_DIST_TYPE

public static final java.lang.String TAG_DIST_TYPE

See Also:

Constant Field Values

TAG_MEAN

public static final java.lang.String TAG_MEAN

See Also:

Constant Field Values

TAG_COVARIANCE

public static final java.lang.String TAG_COVARIANCE

See Also:

Constant Field Values

TAG_PROBABILITIES

public static final java.lang.String TAG_PROBABILITIES

See Also:

Constant Field Values

Constructor Detail

HiddenMarkovModel

public HiddenMarkovModel(int states)

Construct a discrete HMM with the specified number of states.

Parameters:

states - The number of states.

HiddenMarkovModel

public HiddenMarkovModel(int theStates,
                         int theItems)

HiddenMarkovModel

public HiddenMarkovModel(int theStates,
                         int[] theItems)

Method Detail

clone

public HiddenMarkovModel clone()
                        throws java.lang.CloneNotSupportedException

Overrides:

clone in class java.lang.Object

Throws:

java.lang.CloneNotSupportedException

cloneStructure

public HiddenMarkovModel cloneStructure()

createNewDistribution

public StateDistribution createNewDistribution()

getPi

public double getPi(int i)

getStateCount

public int getStateCount()

getStateDistribution

public StateDistribution getStateDistribution(int i)

getStatesForSequence

public int[] getStatesForSequence(MLDataSet seq)

Description copied from interface: MLStateSequence

Get the sates for the given sequence.

Specified by:

getStatesForSequence in interface MLStateSequence

Parameters:

seq - The sequence.

Returns:

The states.

getTransitionProbability

public double getTransitionProbability(int i,
                                       int j)

isContinuous

public boolean isContinuous()

isDiscrete

public boolean isDiscrete()

lnProbability

public double lnProbability(MLDataSet seq)

probability

public double probability(MLDataSet seq)

Description copied from interface: MLStateSequence

Determine the probability of the specified sequence.

Specified by:

probability in interface MLStateSequence

Parameters:

seq - The sequence.

Returns:

The probability.

probability

public double probability(MLDataSet seq,
                          int[] states)

Description copied from interface: MLStateSequence

Determine the probability for the specified sequence and states.

Specified by:

probability in interface MLStateSequence

Parameters:

seq - The sequence.

states - The states.

Returns:

The probability.

setPi

public void setPi(int i,
                  double value)

setStateDistribution

public void setStateDistribution(int i,
                                 StateDistribution dist)

setTransitionProbability

public void setTransitionProbability(int i,
                                     int j,
                                     double value)

updateProperties

public void updateProperties()

Description copied from interface: MLProperties

Update any objeccts when a property changes.

Specified by:

updateProperties in interface MLProperties

Specified by:

updateProperties in class BasicML

getItems

public int[] getItems()

getPi

public double[] getPi()

getTransitionProbability

public double[][] getTransitionProbability()

setTransitionProbability

public void setTransitionProbability(double[][] data)

setPi

public void setPi(double[] data)