PenGE-package: Penalised Gibbs Estimation

In antiphon/PenGE: Penalised Gibbs Estimation

Description

Fit multivariate point pattern Gibbs models using penalised pseudo-likelihoods.

Details

Estimation of interactions in spatial point patterns using intra- and inter-type interacting Gibbs models.

Model inference is based on pseudo-likelihood approximations.

Significance of interactions, and therefore the presence of them, is determined by penalised maximum pseudo-likelihood.

Default penalisation is Group Lasso. Grouped penalisation allows us to determine if the whole multi-scale potentials, which are given by step-functions/basis-functions, are 0 or not.

antiphon/PenGE documentation built on July 31, 2019, 10:01 p.m.