Multivariate peaks over thresholds models (Q1744179)

The manuscript deals with the multivariate peaks over thresholds (PoT) modeling based on generalized Pareto distributions. PoT is one of the most promising methodology for feature selection especially for the extreme episodes prediction. This paper is the first paper where a general point process model is derived for extreme episodes in multi-dimensional situations. The authors have demonstrated how conditioning the distribution of extreme episodes on threshold exceedance gives representations of the family of generalized Pareto distributions. Two illustrative examples are included.