Experiments to improve forecasting accuracy of regression models with minimal assumptions

Abstract

The forecasting accuracy of a regression model relies heavily on the applicability of the assumptions made by the model builder. In addition, the presence of outliers may also lead to models that are not reliable and thus less robust. In this paper a suggested regression model, based on minimal assumptions, is studied and extended in an effort to improve forecast accuracy. The approach is based on mathematical programming techniques combined with smoothing and piecewise linear techniques. Three cases from the literature are considered and presented as illustrative examples.