A Multi-model Dynamic Fusion Forecasting Method and its Application

To overcome the disadvantage of conventional fusion forecasting models, two main research ways are attempted. Firstly, the idea of the comprehensive evaluation is adopted to evaluate the forecasting models which needed to be selected by the multi-attribute comprehensive evaluation model. Secondly, the forecast precision of each forecasting model influenced by the time is compared in rationally way. Thereby, the dynamic fusion forecasting method in which the objectivity and the subjectivity information is integrated is established. Finally, the civil automobile population forecasting problem is taken as research background, the multi-model dynamic fusion forecasting method is put into application, and the application results shows that the validity of the method is validated.