repicea.stats.model.glm

Class GeneralizedLinearModel

java.lang.Object

repicea.stats.model.AbstractStatisticalModel<HierarchicalStatisticalDataStructure>

repicea.stats.model.glm.GeneralizedLinearModel

All Implemented Interfaces:

StatisticalModel<HierarchicalStatisticalDataStructure>

Direct Known Subclasses:

FGMCopulaGLModel

public class GeneralizedLinearModel
extends AbstractStatisticalModel<HierarchicalStatisticalDataStructure>

This class implements generalized linear models.

Field Summary

Fields

Modifier and Type	Field and Description
protected IndividualLogLikelihood	individualLLK
protected LinkFunction.Type	linkFunctionType
protected Matrix	matrixX
protected Matrix	y

Fields inherited from class repicea.stats.model.AbstractStatisticalModel

completeLLK

Constructor Summary

Constructors

Modifier	Constructor and Description
	GeneralizedLinearModel(DataSet dataSet, LinkFunction.Type linkFunctionType, java.lang.String modelDefinition) Constructor using a vector of 0s as starting values for the parameters
	GeneralizedLinearModel(DataSet dataSet, LinkFunction.Type linkFunctionType, java.lang.String modelDefinition, Matrix startingBeta) General constructor
protected	GeneralizedLinearModel(GeneralizedLinearModel glm) Constructor for derived class.

Method Summary

Modifier and Type	Method and Description
protected HierarchicalStatisticalDataStructure	getDataStructureFromDataSet(DataSet dataSet) This method returns the appropriate StatisticalDataStructure from the dataSet.
Matrix	getLinearPredictions()
LinkFunction.Type	getLinkFunctionType() This method returns the type of the link function.
Matrix	getParameters()
Matrix	getPredicted() This method returns a vector of predicted values.
Matrix	getResiduals() This method returns a vector of residuals, that is observed values minus predictions.
protected void	initializeLinkFunction(LinkFunction.Type linkFunctionType)
protected Estimator	instantiateDefaultEstimator() This method defines the default optimizer which is to be specific to the derived classes.
protected void	setCompleteLLK() This method sets the log-likelihood function of the model.
protected void	setModelDefinition(java.lang.String modelDefinition) This method sets the model definition and computes the appropriate matrix from the data.
void	setParameters(Matrix beta)
java.lang.String	toString()

Methods inherited from class repicea.stats.model.AbstractStatisticalModel

doEstimation, getCompleteLogLikelihood, getConvergenceCriterion, getDataStructure, getEstimator, getModelDefinition, getOptimizerParameters, getSummary, setConvergenceCriterion, setOptimizer, setOptimizerParameters

Methods inherited from class java.lang.Object

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

Field Detail

linkFunctionType

protected LinkFunction.Type linkFunctionType

matrixX

protected Matrix matrixX

y

protected Matrix y

individualLLK

protected IndividualLogLikelihood individualLLK

Constructor Detail

GeneralizedLinearModel

public GeneralizedLinearModel(DataSet dataSet,
                              LinkFunction.Type linkFunctionType,
                              java.lang.String modelDefinition,
                              Matrix startingBeta)

General constructor

Parameters:

dataSet - the fitting data

linkFunctionType - the type of ling function (Logit, CLogLog, ...)

modelDefinition - a String that defines the dependent variable and the effects of the model

startingBeta - the starting values of the parameters

GeneralizedLinearModel

public GeneralizedLinearModel(DataSet dataSet,
                              LinkFunction.Type linkFunctionType,
                              java.lang.String modelDefinition)

Constructor using a vector of 0s as starting values for the parameters

Parameters:

dataSet - the fitting data

linkFunctionType - the type of ling function (Logit, CLogLog, ...)

modelDefinition - a String that defines the dependent variable and the effects of the model

GeneralizedLinearModel

protected GeneralizedLinearModel(GeneralizedLinearModel glm)

Constructor for derived class.

Method Detail

getLinkFunctionType

public LinkFunction.Type getLinkFunctionType()

This method returns the type of the link function.

Returns:

a LinkFunction.Type enum variable

setModelDefinition

protected void setModelDefinition(java.lang.String modelDefinition)
                           throws StatisticalDataException

Description copied from class: AbstractStatisticalModel

This method sets the model definition and computes the appropriate matrix from the data.

Overrides:

setModelDefinition in class AbstractStatisticalModel<HierarchicalStatisticalDataStructure>

Parameters:

modelDefinition - a String that defines the model

Throws:

StatisticalDataException

initializeLinkFunction

protected void initializeLinkFunction(LinkFunction.Type linkFunctionType)

setParameters

public void setParameters(Matrix beta)

getParameters

public Matrix getParameters()

instantiateDefaultEstimator

protected Estimator instantiateDefaultEstimator()

Description copied from class: AbstractStatisticalModel

This method defines the default optimizer which is to be specific to the derived classes.

Specified by:

instantiateDefaultEstimator in class AbstractStatisticalModel<HierarchicalStatisticalDataStructure>

Returns:

an Optimizer instance

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getLinearPredictions

public Matrix getLinearPredictions()

getPredicted

public Matrix getPredicted()

Description copied from interface: StatisticalModel

This method returns a vector of predicted values.

Returns:

a Matrix instance

getResiduals

public Matrix getResiduals()

Description copied from interface: StatisticalModel

This method returns a vector of residuals, that is observed values minus predictions.

Returns:

a Matrix instance

getDataStructureFromDataSet

protected HierarchicalStatisticalDataStructure getDataStructureFromDataSet(DataSet dataSet)

Description copied from class: AbstractStatisticalModel

This method returns the appropriate StatisticalDataStructure from the dataSet. It is the first instruction to be called in the constructor.

Specified by:

getDataStructureFromDataSet in class AbstractStatisticalModel<HierarchicalStatisticalDataStructure>

Parameters:

dataSet - a DataSet instance

Returns:

a StatisticalDataStructure derived instance

setCompleteLLK

protected void setCompleteLLK()

Description copied from class: AbstractStatisticalModel

This method sets the log-likelihood function of the model. It is to be defined in the derived class, since the log-likelihood function depends on the different features of the model.

Specified by:

setCompleteLLK in class AbstractStatisticalModel<HierarchicalStatisticalDataStructure>