Event-Driven Competing Risks


Bland Ewing, Brian S. Yandell, James F. Barbieri, and Robert F. Luck
Technical Report #1032, January 2001, U WI Madison Statistics

The non-homogeneous Poisson process with a competing risk structure can be used to simulate the interacting stochastic lives of individuals in an ecological community. Such a technique can help quantify the relationships among observed behaviors of individuals and describe the resulting coupling between interacting populations defined in state space descriptions commonly used in population biology. The event structure provides the dynamics that drives time, rather than the usual time-driven stochastic dynamic programming. We illustrate the ideas with the California red scale-Aphytis host-parasitoid system, although the method has wider applicability.

Click to get manuscript