Entropy method and application to determine weights of combination forecasting

Abstract

Combination forecasting can improve forecasting precision effectively, so it can be utilized to forecast in many fields. The key problem to be solved is how to obtain weighted averaging coefficients of combination forecasting. From point of information theory, the deviation of forecasting error series is defined renewal in this paper, based on the concept of entropy. Then weights of combination forecasting are calculated by this entropy method. This method has clear concept and simple calculation. An example is illustrated to show satisfactory result.