| Contents |
| Self-exciting point processes have shown promise for modeling social event patterns where the occurrence of an event increases the likelihood of subsequent events. However, standard models assume a Poisson background rate for spontaneous events, an unrealistic assumption in many social systems. We introduce a self-exciting Cox process model where the background rate is driven by an Ornstein-Uhlenbeck stochastic differential equation. We then develop a methodology for simultaneous filtering and estimation of the intensity. Application to crime and security data sets are investigated. |
|