Package edu.cmu.tetrad.bayes

Class FactoredBayesStructuralEM

java.lang.Object

edu.cmu.tetrad.bayes.FactoredBayesStructuralEM

public final class FactoredBayesStructuralEM
extends Object

Implements the procedure Factored-Bayesian-SEM found on page 6 of "The Bayesian Structural EM Algorithm" by Nir Friedman.> 0

The initial implementation simplifies the algorithm somewhat by computing the score of each model by using the BdeMetric score, which is implemented in the bayes package of Tetrad-4.2.> 0

Author:

Frank Wimberly, Robert Tillman (changes since 5-12-2008)

Constructor Summary

Constructors

Constructor	Description
FactoredBayesStructuralEM(DataSet dataSet, BayesPm bayesPmM0)

Method Summary

Modifier and Type	Method	Description
DataSet	getDataSet()
BayesIm	iterate()	This iterate2 method also uses BdeMetricCache but it uses the factorScoreMD method which can handle missing data and latent variables.
BayesIm	maximization(double tolerance)	This method allows specification of the tolerance parameter used in Bayes EM estimation.
void	scoreTest()

Methods inherited from class java.lang.Object

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

Constructor Details

FactoredBayesStructuralEM

public FactoredBayesStructuralEM(DataSet dataSet,
 BayesPm bayesPmM0)

Method Details

maximization

public BayesIm maximization(double tolerance)

This method allows specification of the tolerance parameter used in Bayes EM estimation.

iterate

public BayesIm iterate()

This iterate2 method also uses BdeMetricCache but it uses the factorScoreMD method which can handle missing data and latent variables. Ths method iteratively score models and finds that which contains the graph of the highest scoring model (via its BaysPm) as well as parameters which yield the best score given the dataset by using the EmBayesEstimator class.

Returns:

the instantiated Bayes net (BayesIm)

scoreTest

public void scoreTest()

getDataSet

public DataSet getDataSet()