Skip to Main content Skip to Navigation
Theses

Réduction de dimension en présence de données censurées

Abstract : We consider regression models with randomly right-censored responses. We propose new estimators of the regression function in parametric models, and nonparametric lack-of-fit tests of these models. We then adapt these methods to the study of a semiparametric single-index model, in order to generalize dimension reduction techniques used in absence of censoring. We first consider models relying on more restrictive identifiability conditions, and then consider the case when the response and the censoring variable are independent conditionally to the covariates. In this last kind of models, actual techniques do not allow to estimate the regression function when there is more than one covariate. We develop a new dimension reduction approach to circumvent this problem.
Document type :
Theses
Complete list of metadatas

Cited literature [102 references]  Display  Hide  Download

https://pastel.archives-ouvertes.fr/tel-00195261
Contributor : Olivier Lopez <>
Submitted on : Monday, December 10, 2007 - 1:15:28 PM
Last modification on : Friday, July 10, 2020 - 4:04:26 PM
Long-term archiving on: : Thursday, September 27, 2012 - 11:10:15 AM

Identifiers

  • HAL Id : tel-00195261, version 1

Citation

Olivier Lopez. Réduction de dimension en présence de données censurées. Mathématiques [math]. ENSAE ParisTech, 2007. Français. ⟨tel-00195261⟩

Share

Metrics

Record views

1289

Files downloads

1273