Local financial revenue forecasting based on grey theory and neural network

Due to the influence of local financial revenue affected by indeterminate and redundant factors, the nonlinear essence of economic system, this paper attempts to develop a better prediction model by integration of grey theory and neural network techniques. In this research, the grey relationship analysis was used to choose the most important component items for local financial revenue. The grey forecast model, GM (1, 1), was applied to predict each of the reduced items. Finally, a Neural Network was developed to train and predict, with the predicted result of the reduced items as input and the actual data of relevant local financial revenue as output. Results indicated that the grey relationship analysis was able to filter the redundant items. the grey forecast model was able to weaken the fluctuation of data sequence. The neural network was capable of recognizing the relationships between diverse variables. The results show the proposed approach can provide good prediction for the local financial revenue forecasting.