Model selection for hazard rate estimation in presence of censoring

Sandra Plancade (1)

1 MAP5 - UMR 8145 - Mathématiques Appliquées Paris 5

This note presents an estimator of the hazard rate function based on right censored data. A collection of estimators is built from a regression-type contrast, in a general collection of linear models. Then, a penalised model selection procedure provides an estimator which satisfies an oracle inequality. In particular, we can prove that it is adaptive in the minimax sense on Hölder spaces.

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

Sandra Plancade. Model selection for hazard rate estimation in presence of censoring. Metrika, 2011, 74 (3), pp.313-347. ⟨10.1007/s00184-010-0305-9⟩. ⟨hal-00718495⟩