Abstract

Ordinary Least Squares regression is frequently used to model linear data. However, the relationship between variables is not always linear. When faced with this predicament, researchers can use nonparametric regression methods. This paper uses two methods: Nadaraya Watson Estimator (NWE) and Local Linear (LL) Regression. Since prediction is an essential aspect of modeling, simulation studies varying the nonlinear data-generating processes and sample sizes were done to test the predictive abilities of the regression methods. Each regression method's respective mean absolute percentage error (MAPE) was calculated and compared. Results show that the predictive abilities of NWE and LL are comparable, and both are superior to OLS when the data are nonlinear.