Skip to contents

Akaike's Information Criterion and Bayesian Information Criterion for a Poisson-GP Object

Source: R/poisGP.R
AIC.poisGP.Rd

Akaike's Information Criterion and Schwarz's Bayesian Information Criterion for a Poisson-GP model object.

Usage

# S3 method for poisGP
AIC(object, ..., k)

# S3 method for poisGP
BIC(object, ...)

# S3 method for poisGP
BIC(object, ...)

Arguments

object

An object with class "poisGP".

...

Not used yet.

k

See AIC.

Value

Value of the criterion

Note

For technical reasons these methods do not have a type argument as does logLik.poisGP and consequently for a model fitted from block maxima or \(r\)-largest the computed criteria will differ from those computed by other packages because the log-likelihoods differ by a constant.

Caution

Comparing log-likelihoods or related indicators such AIC, BIC across R packages can be irrelevant due to the use of different constants. Moreover the concept of number of observations is unclear when heterogeneous data are used. An historical MAX or OTS information can have a very strong influence on the estimation hence can not be compared to an ordinary OT observation.

See also

logLik.poisGP