Publicación: BAYESIAN ESTIMATION OF THE SHRINKAGE PARAMETER IN RIDGE REGRESSION
Fecha
2020
Título de la revista
ISSN de la revista
Título del volumen
Editor
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
Resumen
A COMMON PROBLEM IN THE PRACTICE OF REGRESSION ANALYSIS IS MULTICOLLINEARITY. ITS NEGATIVE EFFECTS ON THE LEAST SQUARES ESTIMATOR ARE WELL KNOWN. A NUMBER OF SHRINKAGE ESTIMATORS HAVE BEEN DEVELOPED TO CORRECT THIS PROBLEM, PROMINENT AMONG THEM IS RIDGE REGRESSION. WE DISCUSS THE BAYESIAN INTERPRETATION OF THE ESTIMATOR AND IN PARTICULAR WE PROPOSE A BAYESIAN STRATEGY TO SELECT THE SHRINKAGE PARAMETER. WE PROVIDE AN EXAMPLE OF THE RESULTING RIDGE REGRESSION ESTIMATOR. ALSO, TO COMPARE THE ESTIMATION AND PREDICTION PROPERTIES OF THE BAYESIAN SHRINKAGE RIDGE REGRESSION WITH OTHER WELL KNOWN RIDGE REGRESSION AND ORDINARY LEAST SQUARES ESTIMATORS, WE PERFORM A SIMULATION STUDY.
Descripción
Palabras clave
SIMULATIONS, RIDGE REGRESSION, MULTICOLLINEARITY, LINEAR REGRESSION MODEL, BAYESIAN ESTIMATION
