Comparative utilizations of Information Criteria for Gaussian regression on a random design.

Guilhem Coq (1)

1 LMA [Poitiers] - Laboratoire de mathématiques et applications [UMR 6086]

We consider the problem of estimating an unknown function in the setting of Gaussian regression on a random design. To this end, we use general Information Criteria, also called penalized likelihood criteria. We introduce several comparative methods of use of those criteria that present the advantage to have reasonnable computational complexity. We also show that those methods are as efficient as classical ones since they satisfy good asymptotic properties as well as an oracle inequality.

Mots clés: Information Criteria; model selection; Gaussian regression; oracle inequality

Domaines: Statistiques [math.ST]; Théorie [stat.TH]

Guilhem Coq. Comparative utilizations of Information Criteria for Gaussian regression on a random design.. 2008. ⟨hal-00367551v2⟩