Dear All,
We have the pleasure thanks to the support of the ESSEC IDS dpt, Institut des Actuaires, LabEx MMEDII, the group BFA (SFdS), to invite you to the seminar by:
Prof. Antoine USSEGLIO CARLEVE
Département Statistique et Informatique Décisionnelle
Université d'Avignon, France
Date: Wednesday, May 11 at 12:30 pm (Paris) and 6:30 pm (Singapore)
Dual format: ESSEC Paris La Défense (CNIT), Amphi 201, and via Zoom, please click here
( Password/Code : WGRisk)
« Some extreme regression models » If estimating the median (quantile of level 0.5) or the (quantile of level 0.25 or 0.75) of a random variable Y is obvious when we have a sample of size n, what happens if the quantile level exceeds 1/n ? In such a case, the use of the classical order statistic systematically returns the maximum of the sample, and thus leads to a nonconsistent estimation of the quantile. Using extreme value theory, we find in the literature some to extrapolate such extreme quantiles. The particularity of this work is that a covariable vector X is recorded alongside Y . The aim is thus to estimate extreme quantiles of Y given X = x. For that purpose, we firstly propose a purely nonparametric approach, by giving extreme quantile and expectile (an alternative of the classical quantile we will introduce) estimators, and studying their asymptotic properties. The speed of convergence of these estimators being very slow when the size of the covariate X increases, we then propose some dependence models over X and Y, in order to avoid the curse of dimensionality. Some applications in insurance or natural disaster are proposed. Kind regards,
Jeremy Heng, Olga Klopp and Marie Kratz
http://crear.essec.edu/workinggrouponrisk
and Riada Djebbar (Singapore Actuarial Society  ERM)
