Create a function for computation of conditional likelihood

Create a function for the computation of the conditional likelihood of MixAR models for a given time series. The methods for this generic function defined in package mixAR are described here.

Usage

make_fcond_lik(model, ts)

Arguments

model: a MixAR model
ts: the time series

Details

The returned value is a function, say f(x), whose only argument is a numeric vector of parameters with the arrangement of lik_params, for which it computes the conditional loglikelihood. f can be given to optimisation routines.

Argument model is an object inheriting from MixAR and determines the structure of the MixAR model for the function, f, that it creates. So, properties of the model, such as number of components, AR order, and distribution of the noise components are fixed when f is created and only the numeric values of the parameters are changed by calls to it.

Value

a function of one argument, the parameters of a MixAR model as a numeric vector with the arrangement of lik_params, for which it computes the conditional loglikelihood

Todo

The environment of the returned function contains the time series and the model object (initially argument model, later the model used in the last call to f). So, these things can be extracted from f. Is it necessary to create convenience functions?

Methods

signature(model = "MixAR", ts = "numeric")