Prediction And Analysis Of Financial Revenue Of Shenzhen Based On Variable Selection And Grey Neural Network

Shenzhen is one of the central cities of China's economy,the development of local economy in which has a huge role in promoting the development of the national economy.Shenzhen's GDP and high growth rate show that its economic development prospects are quite optimistic.For the financial department of Shenzhen,it is the first condition to solve the financial distribution and optimize the allocation that accurately and effectively forecast and analyze local fiscal revenue.At the same time,it can provide support for the relevant tax policies.For other local government departments,understanding and mastering the local economic structure and industrial structure of Shenzhen has a certain enlightening effect on their own industrial upgrading and transformation.China has implemented the fiscal tax sharing system since 1994.Local finance is not only an important part of national income,but also has an independent structure.Local fiscal revenue can reflect the local economic operation,and good local fiscal revenue can contribute to the social economy.While ensuring residents' living standards and improving residents'income,it provides direction and rectification suggestions for underdeveloped areas.This paper selects the data of Shenzhen's financial revenue and 27 influencing factors since 1994,Firstly,descriptive statistics and correlation analysis are carried out to understand the distribution characteristics of the data.Then,Lasso method is used to estimate the variable coefficients.Combined with 10 fold cross-validation method,the original data is divided into training set and validation set.Using the minimum angle regression algorithm to solve the estimation value,so that some variable coefficients are 0,and at the same time,it can produce estimation value quickly and accurately.Therefore,7 main explanatory variables are obtained by eliminating some collinearity and less influential variables,which are general public budget expenditure,total retail sales of social consumer goods,total value of the tertiary industry,ratio of tertiary industry to secondary industry,total agricultural output value,permanent population at the end of the year and total wages of on-the-job employees.Considering that all the variables of the model are time series data,and the model dimension is high but the sample size is not large,the grey model GM(1,1)is established for seven main explanatory variables,and the grey prediction is carried out to give play to the poor information processing ability of the grey model.The predicted values of the main explanatory variables in 2019 and 2020 are calculated,and the prediction accuracy is tested whose result is positive.Considering that the artificial neural network has good processing and self-learning capabilities for nonlinear complex models.The grey prediction model is combined with BP neural network for combination prediction,the neural network is trained with the data of the first 25 years,the gray prediction value is brought into the trained neural network model,and the financial prediction value of Shenzhen in 2 019 and 2020 is calculated.