4-8 Mar 2013 Cotonou (Benin)
Wednesday 6
Statistics 2

› 18:00 - 18:30 (30min)
› Salle de conférence 1
Estimation non-paramétrique des quantiles extrêmes conditionnels en présence de censure.
Pathé Ndao  1@  , Aliou Diop  2@  , Jean-François Dupuy@
1 : LERSTAD
Campus Social Université Gaston Berger de Saint-Louis N° chambre: 61 L BP: 234 Saint-Louis Sénégal -  Sénégal
2 : LERSTAD Universite Gaston Berger  (LERSTAD)

In this paper, we consider the problem of estimating the conditional extreme value index and extreme conditional quantiles in the presence of censorship. Einmahl et al. (2008) proposed estimators of the extreme value index and quantile extremes in the case of censored data without covariate. In this work, we propose adaptations of these estimators in the case of censored randomly to the right and the presence of covariates. Then a theoretical study of asymptotic properties of these new estimators will be proposed. Finally, their behavior is illustrated on the basis of simulations.



  • Presentation
Online user: 1