Class "response"

Description

A generic response model for depmix models.

Arguments

object

An object of class "response".

Details

This class offers a framework from which to build specific response models such as glm based responses or multinomial responses. For extensibility, objects with class response should have at least methods: dens to return the dens'ity of responses, and getpars and setpars methods to get and set parameters.

The constr slot is used for information on constraints that are inherently part of a model; the only such constraints which are currently used are 1) the sum constraint in multinomial models with identity link, and 2) a lower bound of zero of sd parameters in gaussian distributions. Such constraints are only used in fitting models with general optimization routines such as Rsolnp; the EM algorithm automagically respects the sum constraint.

lin:

Derivative of linear constraint.

linup:

Upper bounds for linear constraints.

linlow:

Lower bounds for linear constraints.

parup:

Upper bounds on parameters.

parlow:

Lower bounds on parameters.

Slots

parameters:

A (named) list of parameters.

fixed:

A logical vector indicating which parameters are fixed.

y:

A matrix with the actual response; possibly multivariate.

x:

A design matrix; possibly only an intercept term.

npar:

The number of parameters.

constr:

Information on constraints.

Accessor Functions

The following functions should be used for accessing the corresponding slots:

npar:

The number of parameters of the model.

getdf:

The number of non-fixed parameters.

Author(s)

Ingmar Visser & Maarten Speekenbrink