Spatiotemporal point processes: regression, model specifications and future directions (Q2330482)

Inference procedures for point processes are discussed by focusing on computational aspects. In particular, Poisson processes (mainly on discrete time) are considered when the effects of explanatory variables are also taken into account. In such a model framework, the likelihood depends on the intensity function and cannot be computed analytically. To overcome such difficulties, two main approaches are discussed: one is based on discretization (useful when regional effects rather than pointwise effects are important), the other one on augmentation. Two applications illustrate the main procedures. Finally, possible directions of future works are outlined.