Skip to contents

Log-likelihood of a Poisson-GP model object.

Usage

# S3 method for poisGP
logLik(object, type = c("poisGP", "PP"), ...)

Arguments

object

An object with class "poisGP".

type

An experimental argument to make the log-likelihood comparable to that of other packages/functions. See Details.

...

Not used yet.

Value

Value of the log-likelihood.

Details

Although models for block maxima and \(r\)-largest models can be considered as special cases of a Poisson-GP model, the log-likelihood as computed in classical Extreme-Value packages can differ by a constant related to the observation scheme. The argument type aims at finding the same results in the two cases where the results can be obtained with classical packages, namely: "POT" fits with Poisson-GP representation and block maxima or r-largest. In the first case, this is hopefully attained by choosing type to be "poisGP" and in the second case by choosing type to be "PP".

Caution

Comparing log-likelihoods or related indicators such AIC, BIC across R packages can be irrelevant due to the use of different constants.