Methods for creating depmix transition and initial probability models

Share:

Description

Create transInit objects for depmix models using formulae and family objects.

Usage

1
2
3
4
5
6
	
	transInit(formula, nstates, data=NULL, family=multinomial(),
		pstart=NULL, fixed=NULL, prob=TRUE, ...)
		
		## S4 method for signature 'transInit'
getdf(object)

Arguments

formula

A model formula.

data

An optional data.frame to interpret the variables from the formula argument in.

family

A family object; see details.

pstart

Starting values for the coefficients.

fixed

Logical vector indicating which paramters are to be fixed.

prob

Logical indicating whether the starting values for multinomial() family models are probabilities or logistic parameters (see details).

nstates

The number of states of the model.

object

An object of class transInit.

...

Not used currently.

Details

The transInit model provides functionality for the multinomial probabilities of the transitions between states, as well as for the prior or initial probabilities. These probabilities may depend on (time-varying) covariates. The model can be used with link functions mlogit and identity; the latter is the default when no covariates are. With the mlogit link function, the transition probabilities are modeled as baseline logistic multinomials (see Agresti, 2002, p. 272 ff.).

Start values for the parameters may be provided using the pstart argument; these can be provided as probabilities, the default option, or as baseline logistic parameters, use the prob argument to specify the chosen option. The default baseline category is set to 1, which can be modified through calling, say, family=multinomial(base=2).

Note that the transInit model extends the response-class, but that it actually lacks a reponse, i.e. the y-slot is empty, at the time of construction, as the transitions are not observed.

getdf returns the number of free parameters of a transInit model.

Value

transInit return objects of class transInit; this class extends the response-class.

Author(s)

Ingmar Visser & Maarten Speekenbrink

References

Agresti, A. (2002). Categorical Data Analysis. Wiley series in probability and mathematical statistics. Wiley-Interscience, Hoboken, NJ, 2 edition.