A Composite Likelihood Approach in Fitting Spatial Point Process Models

Abstract

We propose a new likelihood-based approach in fitting spatial point process models. A composite likelihood is first formed by adding some pairwise composite likelihood functions that are defined in terms of the second-order intensity function of the underlying process, and then used for estimating the unknown parameters. The estimation procedure is computationally simple and yields consistent and asymptotically normal estimators under some mild conditions. We demonstrate through a simulation study and applications to two real data examples that the proposed approach may lead to improved estimations compared with the commonly used “minimum contrast estimation” approach.