A0206
Title: On extremal cluster inference
Authors: Olivier Wintenberger - Sorbonne Universite (France) [presenting]
Gloria Buritica - AgroParisTech (France)
Abstract: Extremes occur in stationary, regularly varying time series as short periods with several large observations, known as extremal clusters. Cluster statistics are studied, such as the extremal index, cluster size probabilities, and other cluster indices. The asymptotic normality of block estimators is stated for cluster inference based on consecutive observations with large $\ell^\alpha$-norms, where $\alpha$ is the index of regular variation that can be approximated using the Hill estimator. Findings are illustrated on simulations and environmental data.
