Tail dimension reduction for extreme quantile estimation

Laurent Gardes (1)

1 IRMA - Institut de Recherche Mathématique Avancée

In a regression context where a response variable Y ∈ R is recorded with a p-dimensional covariate X , two situations can occur simultaneously in some applications: (a) we are interested in the tail of the conditional distribution and not on the central part of the distribution and (b) the number p of regressors is large. Up to our knowledge, these two situations have only been considered separately in the literature. The aim of this paper is to propose a new dimension reduction approach adapted to the tail of the distribution and to introduce an extreme conditional quantile estimator. A simulation experiment and an illustration on a real data set were presented.

Mots clés: Regression; extreme quantile; dimension reduction; kernel smoothing

Domaines: Statistiques [math.ST]

Laurent Gardes. Tail dimension reduction for extreme quantile estimation. 2016. ⟨hal-01322374v1⟩