A0919
Title: Bayesian nonparametric model of marked Hawkes processes, with application to earthquake occurrences
Authors: Hyotae Kim - Duke University (United States) [presenting]
Athanasios Kottas - University of California Santa Cruz (United States)
Abstract: The aim is to propose a Bayesian nonparametric model for marked Hawkes processes (MHPs). The processes' conditional intensity function is decomposed into the ground process intensity and the mark density function. The primary concentration is modelling the process intensity, but it offers several choices for the mark density. The prior probability model for intensity has been carefully designed to provide model flexibility and tractable posterior inference using a novel mixture modeling method. This model was motivated by seismology applications, where magnitude is regarded as a mark associated with a time point for earthquake occurrence. Accordingly, the mixture model basis is a function of occurrence time and magnitude, with its functional form selected considering not only model flexibility but also earthquake data characteristics, for example, the fact that earthquakes of greater magnitude cause more subsequent shocks than earthquakes of smaller magnitude. The model is illustrated with an earthquake occurrence dataset and several synthetic examples.
