Model Development
In this step, all the collected data are analyzed using PASW Statistics 18. Four selection methods, namely (1) Enter, (2) Forward, (3) Backward, and (4) Stepwise, are adopted to develop the forecasting model based on the multiple regression technique. The forecasting model based on the multiple regression technique provides a set of independent variables which also present an estimated score on each dependent variable. From this step, the formulation of the forecasting model is performed in the form of Equation (1), as addressed previously. Model Verification and Validation Among the various forecasting models with different selection methods for multiple regression models, MAPE comparison is conducted to finalize the selection of a practical model. The best-fit forecasting model is selected from among the four selection methods by considering the values of the mean absolute percent error (MAPE) which is calculated using the formula given in Equation (2):