The smoothing parameter selection problem in smoothing spline regression for different data sets

Abstract

This paper studies smoothing parameter selection problem in nonparametric regression based on smoothing spline method for different data sets. For this aim, a Monte Carlo simulation study was performed. This simulation study provides a comparison of the five popular selection criteria called as cross-validation (CV), generalized cross-validation (GCV), improved Akaike information criterion (AICc), Mallows' Cp and risk estimation using classical pilots (RCP). Empirical performances of five selection criteria were examined for this simulation and the most suitable one was selected accordingly.