Cross Validation for the Log Gaussian Cox process

Djidenou (Hans) Montcho (KAUST)

Wednesday 1st October 11:00-12:00
Maths 116

Abstract

The Log Gaussian Cox process (LGCP) is arguably one of the most used models to analyze spatial point pattern (SPP) data due to its flexibility to handle heterogeneity in the underlying latent process. The literature on fitting such models is relatively mature both under a frequentist and Bayesian framework, however practical tools for model assessment and comparison have not seen the same development.  This work provides, under a Bayesian framework, a theoretically justified and more importantly fast, scalable and practical solution to some of these problems using Cross Validation.

Add to your calendar

Download event information as iCalendar file (only this event)