A simulation study with different number of observation using nonparametric regression splines

Abstract

In this study we made a simulation study using various nonparametric techniques. The methods that we have used in this study are following: regression spline, penalized spline, and their Bayesian versions: adaptive Bayesian regression spline and Bayesian P-splines. The main goal of our study is to compare nonparametric regression techniques and Bayesian versions of these techniques. For this purpose we made a simulation study with different functions. For each function we sampled n = 50, n =100 n = 200 n = 400 number of observations. The purpose of using different number of sampled observations is to analyze the behavior of utilized techniques. For simulation study we used dataset which sampled from the functions in the paper of [7]. The results of simulation study are compared with each other using mean value of the MSE (mean squared error) and showed results graphically using box plot of MSE.