Properties of weighted geometric means combination forecasting model based on degree of logarithm grey incidence

We build the weighted geometric means combination forecasting model based on degree of logarithm gray incidence. This is a new kind of nonlinear combination forecasting method. We calculate weighted coefficient vectors of combination forecasting methods by the optimal criteria of maximizing the logarithm the degree of gray incidence of combination forecasting model, which is different from ordinary criteria of minimizing absolute errors. We put forward some new concepts of this model, such as superior combination forecasting, redundant degree of combination forecasting etc. We exhibit the sufficient conditions of the existence of non-inferior and superior combination forecasting. We also prove that there exists redundant single forecasting method in the combination forecasting if the single forecasting method is strictly dominant over another. These results show that the combination forecasting model is effective theoretically.