Fitting a linear regression model by combining least squares and least absolute value estimation

Robust estimation of the multiple regression is modeled by using a convex combination of Least Squares and Least Absolute Value criterions. A Bicriterion Parametric algorithm is developed for computing the corresponding estimates. The proposed procedure should be specially useful when outliers are e...

ver descrição completa

Detalhes bibliográficos
Autores: Allende, Sira, Bouza, Carlos, Romero, Isidro
Tipo de documento: artigo
Data de publicação:1995
País:España
Recursos:Universitat Politècnica de Catalunya (UPC)
Repositório:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglês
OAI Identifier:oai:upcommons.upc.edu:2099/4056
Acesso em linha:https://hdl.handle.net/2099/4056
Access Level:Acceso aberto
Palavra-chave:Inference
Outliers in regression
L1 regression
Bicriterion parametric algorithm
Inferència
Classificació AMS::62 Statistics::62F Parametric inference
Classificació AMS::62 Statistics::62J Linear inference, regression
Descrição
Resumo:Robust estimation of the multiple regression is modeled by using a convex combination of Least Squares and Least Absolute Value criterions. A Bicriterion Parametric algorithm is developed for computing the corresponding estimates. The proposed procedure should be specially useful when outliers are expected. Its behavior is analyzed using some examples.