Application and evaluation on combination forecasting model based on information entropy and Shapley value

The key problem of combination forecasting model is the weight of the single prediction methods. The values of the weights directly effect the accuracy of combination forecasting model. In this paper, taking agricultural machine total power in Heilongjiang province as original data, the three combination forecasting models were obtained through respectively using shapley value, information entropy and combining information entropy with shapley value to calculate weights of each single models. The combination forcasting model based on combining shapley value with information entropy after evaluation could get the highest prediction accuracy.