The Construction Of China's Local Government Debt Risk Identification System

In recent years,with the gradual increase of economic downward pressure and gradual slowdown of provincial fiscal revenue growth,the debt problems of many local governments begin to appear.In the future,some local governments will be wary of de facto bankruptcy.Therefore,It is necessary to carry out proactive research on the debt problem of local governments.The debt risk of local governments is not a sudden phenomenon,but a gradual process of change.There must be some signals before the crisis of risk.This provides the forecast risk may.There are also many factors that affect local government debt risk,and there are also complex nonlinear relationships among these factors.Therefore,it is significant to identify the debt risk of local governments in China by using a theoretical model that can describe nonlinear relationships and build an objective and comprehensive risk assessment system.This paper mainly uses the improved decision tree model based on C4.5algorithm to carry out risk identification of local debt.Due to the difficulty of getting local government debt in our country and the incompleteness of many data,we can only estimate it.In this paper,the basic data of local government debt are estimated from the perspective of government revenue and expenditure to calculate the risk of local government debt Related indicators,and then use the factor analysis to do the reliability test to eliminate some unnecessary indicators,and draw the comprehensive index value,the use of integrated index value of China's local debt risk is divided into five categories,namely,risk-free,normal,mild risk,moderate risk and severe risk.Finally,the indicator data of factor analysis and the results obtained are input into the C4.5 algorithm-based decision tree model for comprehensive evaluation.As a typical data mining model,the decision tree model has the advantages of wide application,easy understanding and high classification accuracy.However,the original decision tree model generally treats all the misclassifications as equals,which makes the model's cost value in the misclassification higher.To solve this problem,this paper introduces the cost matrix and combines the different types of misjudgment costvalues.Cost matrix to achieve optimization of the C4.5 algorithm.The conclusion shows that in many provinces,there are indeed obvious risk issues in recent years.At the same time,the improved decision tree model based on C4.5 algorithm can effectively reduce the misjudgment cost of the predictive classification model,and it is also accurate in predicting the risk of local debt.It is obviously better than other models,indicating that this model can better identify local government debt risk.