Author(s)

Zhongyao Wang, Dianya Pan, Jia Yang

Corresponding Author

Zhongyao Wang

ABSTRACT

The standard uses deep confidence network deep learning algorithm to study the short-term prediction of local economy. Firstly, the strong correlation index of local economic prediction and analysis is extracted, and the index data are obtained quickly and effectively through the Internet by crawler program. With the help of Python platform, the index is output automatically, and the output data are normalized and vectorized. Taking the local GDP as the prediction object, the short-term forecast of local economy is completed through DBN deep learning algorithm. In the process of DBN training, the weight parameters are cross-adjusted forward and backward, and the optimal parameters are obtained to determine the DBN structure. Taking Chongqing as an example, the simulation results show that the short-term prediction accuracy of local economy based on big data drive and DBN deep learning is high and the MSE is small. The experimental results show that the DBN deep learning algorithm has higher accuracy and smaller MSE in predicting local GDP compared with common economic prediction algorithms. The following research will further study DBN network scale, learning rate and other aspects, in order to improve the time performance of local economic short-term prediction.

KEYWORDS

Deep Learning, Economic Forecast, Deep confidence Network, Gross domestic Product