Parametric Additive and Semiparametric Models in the Presence of Heteroscadicity

• Activity Number: 676
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Ahmet Sezer*+
• Companies: Anadolu University
• Keywords: Semiparametric ; Polynomial ; Spline

Abstract

Parametric and semiparametric additive models are tools with a wide range of applications. Polynomial model is linear in the coefficients even though it is a nonlinear function of the predictor variable. Like polynomial models, in many disciplines there are theoretical models relating the predictors and the response and often these models are nonlinear in their parameters. In many cases, semiparametric models are alternatively used over the polynomial and nonlinear models for the nonlinear relationships. In this study, we deal with when to use semiparametric models(splines) over the parametric additive models. A simulation study is conducted to compare these two different approaches by using the mean average squared error (MASE). Our results indicate that presence of heterocasdicity and sample size affects MASE.