Smoothing Spline Semiparametric Regression Model Assumption Using PWLS Approach

Abstract

Semiparametric regression analysis is a combination of parametric regression and nonparametric regression. Semiparametric regression analysis arises because there are cases of modeling where the relationships between variables other than linear are also unknown in the form of the regression curve. Semiparametric regression using smoothing spline is one tool that has high flexibility and discusses the whole form of regression. The purpose of this study was to apply semiparametric smoothing spline regression analysis using the Penalized Weighted Least Square (PWLS) approach to the intention variable to use E-BISTER innovation. This study uses a questionnaire as a tool to collect data. Many respondents involved were determined using proportional sampling method with accidental techniques, namely as many as 96 active students of the Faculty of MIPA Universitas Brawijaya. The results of the study show that the semiparametric smoothing spline regression model using PWLS is a good predictor of the model for this case because the diversity of data that can be explained by the model is 65.93% while the remaining 34.07% is influenced by other variables.