18
14

Discrete Extremes

Abstract

Our contribution is to widen the scope of extreme value analysis applied to discrete-valued data. Extreme values of a random variable XX are commonly modeled using the generalized Pareto distribution, a method that often gives good results in practice. When XX is discrete, we propose two other methods using a discrete generalized Pareto and a generalized Zipf distribution respectively. Both are theoretically motivated and we show that they perform well in estimating rare events in several simulated and real data cases such as word frequency, tornado outbreaks and multiple births.

View on arXiv
Comments on this paper