org.encog.ml.bayesian
Class BayesianEvent

java.lang.Object
  org.encog.ml.bayesian.BayesianEvent

All Implemented Interfaces:

Serializable

public class BayesianEvent
extends Object
implements Serializable

Events make up a Bayesian network. Each evidence or outcome event usually corresponds to one number in the training data. A event is always discrete. However, continues values can be range-mapped to discrete values.

See Also:

Serialized Form

Constructor Summary

BayesianEvent(String theLabel)
Construct a boolean event.

BayesianEvent(String theLabel, List<BayesianChoice> theChoices)
Construct an event with the specified label and choices.

BayesianEvent(String theLabel, String[] theChoices)
Construct an event with the specified label and choices.

Method Summary

void	addChild(BayesianEvent e) Add a child event.
void	addParent(BayesianEvent e) Add a parent event.
int	calculateParameterCount()
void	finalizeStructure() Finalize the structure.
static String	formatEventName(BayesianEvent event, int value) Format the event name with +, - and =.
List<BayesianEvent>	getChildren()
BayesianChoice	getChoice(int arg) Return the choice specified by the index.
Set<BayesianChoice>	getChoices()
String	getLabel()
List<BayesianEvent>	getParents()
BayesianTable	getTable()
boolean	hasChildren()
boolean	hasGiven(String l) Return true if the event has the specified given event.
boolean	hasParents()
boolean	isBoolean()
int	matchChoiceToRange(double d) Match a continuous value to a discrete range.
void	removeAllRelations() Remove all relations.
void	reset() Reset the logic table.
boolean	rollArgs(double[] args) Roll the specified arguments through all of the possible values, return false if we are at the final iteration.
String	toFullString()
String	toString()
void	validate() Validate the event.

Methods inherited from class java.lang.Object

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

Constructor Detail

BayesianEvent

public BayesianEvent(String theLabel,
                     List<BayesianChoice> theChoices)

Construct an event with the specified label and choices.

Parameters:

theLabel - The label.

theChoices - The choices, or states.

BayesianEvent

public BayesianEvent(String theLabel,
                     String[] theChoices)

Construct an event with the specified label and choices.

Parameters:

theLabel - The label.

theChoices - The choices, or states.

BayesianEvent

public BayesianEvent(String theLabel)

Construct a boolean event.

Parameters:

theLabel - The label.

Method Detail

getParents

public List<BayesianEvent> getParents()

Returns:

the parents

getChildren

public List<BayesianEvent> getChildren()

Returns:

the children

getLabel

public String getLabel()

Returns:

the label

addChild

public void addChild(BayesianEvent e)

Add a child event.

Parameters:

e - The child event.

addParent

public void addParent(BayesianEvent e)

Add a parent event.

Parameters:

e - The parent event.

hasParents

public boolean hasParents()

Returns:

True, if this event has parents.

hasChildren

public boolean hasChildren()

Returns:

True, if this event has parents.

toFullString

public String toFullString()

Returns:

A full string that contains all info for this event.

toString

public String toString()

Overrides:

toString in class Object

calculateParameterCount

public int calculateParameterCount()

Returns:

The parameter count.

getChoices

public Set<BayesianChoice> getChoices()

Returns:

the choices

getTable

public BayesianTable getTable()

Returns:

the table

finalizeStructure

public void finalizeStructure()

Finalize the structure.

validate

public void validate()

Validate the event.

isBoolean

public boolean isBoolean()

Returns:

True, if this is a boolean event.

rollArgs

public boolean rollArgs(double[] args)

Roll the specified arguments through all of the possible values, return false if we are at the final iteration. This is used to enumerate through all of the possible argument values of this event.

Parameters:

args - The arguments to enumerate.

Returns:

True if there are more iterations.

removeAllRelations

public void removeAllRelations()

Remove all relations.

formatEventName

public static String formatEventName(BayesianEvent event,
                                     int value)

Format the event name with +, - and =. For example +a or -1, or a=red.

Parameters:

event - The event to format.

value - The value to format for.

Returns:

The formatted name.

hasGiven

public boolean hasGiven(String l)

Return true if the event has the specified given event.

Parameters:

l - The event to check for.

Returns:

True if the event has the specified given.

reset

public void reset()

Reset the logic table.

matchChoiceToRange

public int matchChoiceToRange(double d)

Match a continuous value to a discrete range. This is how floating point numbers can be used as input to a Bayesian network.

Parameters:

d - The continuous value.

Returns:

The range that the value was mapped into.

getChoice

public BayesianChoice getChoice(int arg)

Return the choice specified by the index. This requires searching through a list. Do not call in performance critical areas.

Parameters:

arg - The argument number.

Returns:

The bayesian choice found.