Bandwidth selection for local regression methods have been studied in the literature. However, the classical criteria such as AICc and GCV ignore some information from the estimated coefficients, leading to biased assessment. In this work, we propose utilizing full-information criteria for bandwidth selection and explore their performance through a simulation study in comparison with other existing methods.
Bandwidth selection for local regression methods have been studied in the literature. However, the classical criteria such as AICc and GCV ignore some information from the estimated coefficients, leading to biased assessment. In this work, we propose utilizing full-information criteria for bandwidth selection and explore their performance through a simulation study in comparison with other existing methods.