Prediction And Analysis Of Wuhan's Financial Revenue Based On Data Mining Technology

Local fiscal revenue is a total reflection of regional national economy,and is also the basis for macro-control by market economy countries.Accurate financial revenue prediction is the prerequisite for optimizing the financial budget,leading scientific and accurate financial management.It has a very important practical value for the relevant departments to formulate effective financial policies and budgets,to strengthen the supervision and management of local financial revenue,and to promote the rapid and coordinated development of the national economy.The level of economic development,tax policy,and macro economic environment affect the financial income in varying degrees.The cross and combination of various factors have increased the difficulty of financial forecast.Therefore,in order to deal with more and more complex data,more novel and effective forecasting methods should be proposed.As the largest economic entity in the central region,the economic development of Wuhan city plays an important role in the national economy.This thesis chooses the data of fiscal revenue of Wuhan during 1994 and 2017,and 12 influencing factors.First of all,this thesis dose the correlation and collinearity analysis on the data,from which we find that there is exists some problem of these data.Then the Adaptive-lasso method is used to estimate the variable coefficient,and the Least angle regression is used to solve the Adaptive-lasso estimation.As a result,collinearity and less influential variables are eliminated.Secondly,GM(1,1)Grey-forecasting base on residual variables is used to obtain the predictive value of 2018 and 2019,and then evaluat the grade of prediction accuracy.Finally,this thesis use the data of the previous 24 years to train the neural network,and bring the value of the gray forecast into the trained neural network to get the financial forecast of Wuhan in 2018 and 2019.On the hand,this thesis select 12 influencing factors in a wide range.On the other hand,the data is select from 1994 to 2017,so the time span is large.Therefore,the result isobtain before considering the economic and historical background fully,and it hascertain effect.