Regression Analysis is at the center of almost every Forecasting technique, yet few people are comfortable with the Regression methodology. We hope to improve the level of comfort with this article. In this article we briefly discuss the theory behind the methodology and then outline a step-by-step procedure, which will allow almost everyone to construct a Regression Forecasting function for both the linear and some non-linear cases. Also discussed, in addition to the model construction mentioned above, is model testing (to establish significance) and the procedure by which the Final Regression equation is derived and retained to be used as the Forecasting equation. Hand solutions are derived for some small-sample problems (for both the linear and non-linear cases) and their solutions are compared to the MINITAB-derived solutions to establish confidence in the statistical tool, which can be used exclusively for larger problems.
Category - Athanasios VASILOPOULOS
St. John’s University, United States