Rencontres Statistiques Lyonnaises

Generalized Pareto regression trees for extreme event analysis

by Maud THOMAS (Sorbonne Université)

Europe/Paris
Salle Fokko du Cloux au 1er étage (La doua, bâtiment Braconnier)

Salle Fokko du Cloux au 1er étage

La doua, bâtiment Braconnier

Description

Generalized Pareto regression trees for extreme event analysis

We provide finite sample results to assess the consistency of Generalized Pareto regression trees, as tools to perform extreme value regression. The results that we provide are obtained from concentration inequalities, and are valid for a finite sample size, taking into account a misspecification bias that arises from the use of a ``Peaks over Threshold'' approach. The properties that we derive also legitimate the pruning strategies (i.e. the model selection rules) used to select a proper tree that achieves compromise between bias and variance. The methodology is illustrated through a simulation study, and a real data application in insurance against natural disasters.

joint work with S. Farkas, A. Heranval and O. Lopez.

 

 

A distance : https://univ-lyon1.webex.com/univ-lyon1/j.php?MTID=m5d77c11315452ecb58aa3988d4b08a98