Remarques sur le maximum de vraisemblance

Some paradoxes on the maximum likelihood principle are presented and commented. We consider the properties of the maximum likelihood estimators as a particular case of the M-estimators. We propose a unified theory which includes non-dominated models. Several examples are given.

Detalhes bibliográficos
Autores: Huber, Catherine, Nikulin, M. S.
Formato: artículo
Fecha de publicación:1997
País:España
Recursos:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:francés
OAI Identifier:oai:upcommons.upc.edu:2099/4079
Acesso em linha:https://hdl.handle.net/2099/4079
Access Level:acceso abierto
Palavra-chave:Inference
Vraisemblance
Inferència
Classificació AMS::62 Statistics::62F Parametric inference
Descrição
Resumo:Some paradoxes on the maximum likelihood principle are presented and commented. We consider the properties of the maximum likelihood estimators as a particular case of the M-estimators. We propose a unified theory which includes non-dominated models. Several examples are given.