Multivariate extreme value theory

Anne Sabourin 


This course is an introduction to multivariate extreme value theory and modeling. 

On the probabilistic side, we will introduce  the main ingredients of multivariate extreme value models via the notions of multivariate regular variation and vague convergence,   convergence of empirical measures towards a Poisson process, the exponent and angular measures of a regularly varying random vector, and the existing relationships between the convergence of componentwise maxima of independent multivariate observations and the convergence  of the exceedances over a high threshold towards a Generalized Pareto distribution. 

From a statistical perspective, several modeling strategies for the analysis of  extreme values will be presented, based on block maxima or peaks-over-threshold, together with commonly used parametric or semi-parametric models (logistic family, pairwise Beta model, Dirichlet mixture model). 

Date Course Place
11 Feb 2016 09:30 Multivariate extreme value theory Université Paris Ouest Nanterre, Bâtiment G, Room 614
18 Feb 2016 09:30 Multivariate extreme value theory Université Paris Ouest Nanterre, Bâtiment G, Room 614