compareGP {hetGP}

Likelihood-based comparison of models

Description

Compare two models based on the log-likelihood for hetGP and homGP models

Usage

compareGP(model1, model2)

Arguments

model1, model2

hetGP or homGP models

Value

Best model based on the likelihood, first one in case of a tie

Note

If comparing homoskedastic and heteroskedastic models, the un-penalised likelihood is used for the later, see e.g., (Binois et al. 2017+).

References

M. Binois, Robert B. Gramacy, M. Ludkovski (2018), Practical heteroskedastic Gaussian process modeling for large simulation experiments, Journal of Computational and Graphical Statistics, 27(4), 808–821.
Preprint available on arXiv:1611.05902.