An optimal combination forecasting model based on maximum closeness of TOPSIS and its properties

Technique for order preference by similarity to ideal solution (TOPSIS) is a common decision-making method. Here it was used as a new criterion for forecasting accuracy. Based on the concepts of positive ideal point sequence and negative ideal point sequence, the optimal combination forecasting model was constructed based on the criterion of maximizing closeness degree in the TOPSIS model. Based on this model, some new concepts were defined, such as the non-inferior combination forecasting, superior combination forecasting, redundant forecasting method and dominant forecasting method. Mathematical properties of this model were studied, in which the advantages of optimal combination forecasting were expounded theoretically. Finally, an example was given and was compared with other combined forecasting models. The results show that the combined forecasting model is feasible and effective.