Study Of Tax Revenue Prediction Based On BP Neural Networks And Grey Model

Tax is a kind of allotment relation of the public finance income forcibly and gratisly, which is carried out by a national political power according to the standard that national law enacts. With economic development, it seem increasingly important in making prediction about tax revenue, which determines tax plans and Economic Decision. But because of the effects of tax classification and other related factors, it is one of subjects to be researched continuously to predict tax with great accuracy.Firstly, we introduce the basic conception and method of commonly used prediction models, with emphasis on the analysis of the auto regressive moving average prediction model, Markov chain model and Exponential Smoothing Method, and we propose a tax revenue forecast based on grey model and related research method based on BP neural networks and analyses the principle and process.Bsaed on the above works, prediction program is designed and implemented in this paper, and we finish the simulation testing according to the tax data from Shangluo State Taxation Bureau, Shaanxi province. At last we carry on a contrast analysis of grey model and BP neural networks model and analyze their advantages, shortages and what environments they can be used.