A Novel Grey Model for Multi-regional Macro-data Forecasting by Considering Spatial Correlation and Actual-state Rolling.

Accurate prediction of regional development trend is important to regional planning and coordinated development in China. It provides a basis for decision-makings on the resource balance in multi-regional integration. However, due to the limitation of macro data and the influence of multi-regional correlation, the prediction accuracy of the existing forecasting methods in multi-regional macro-data forecasting is reduced. To overcome these problems, an improved grey model is proposed in this study. Firstly, a new spatial weight matrix is constructed based on the grey correlation analysis to define the spatial effect of multiple regions. Then, an actual-state rolling spatial-effect weighted grey model (ARSWGM) is developed considering the spatial interactions and the actual-state rolling mechanism. Finally, the proposed model is validated by the forecasting of manufacturing quality level of representative provinces in the process of regional coordinated development in China. The result shows that the proposed model demonstrates the best predicting performance compared with the classical grey forecasting models, indicating the advantages of this proposed model in multi-regional macro-data forecasting. Furthermore, this model can also be applied for a broader range of multi-regional limited macro-data forecasting. [ABSTRACT FROM AUTHOR]