Research On Fraud Risk Identification Model Of Listed Companies Based On Random Forest

At present,China is in a critical period of financial system transformation and is gradually becoming an important driving force for global economic growth.However,in the past two decades,the fraudulent scandals of listed companies have been frequent and repeated,and the confidence of investors and the public in the capital market has been severely hit,which has reduced the credibility of the company's financial reports.Whether they can effectively control corporate fraud and the capital market is ill,will determine the capital market and the real economy in the new era.The success of the docking and the efficiency of industrial transformation and upgrading under the structural reform of the supply side are highly concerned by the accounting theory,practice and regulatory agencies.The research shows that the model recognition fraud effect is better than the case analysis.At present,the research on fraud identification indicators is relatively perfect,and the construction of fraud identification model remains to be explored.Based on this paper,the introduction of random forest algorithm to identify fraud in listed companies has important practical significance for maintaining the effective vitality of the capital market.As a combined classifier algorithm,Random Forest can maintain high prediction accuracy on large samples,high dimensional features and outlier data,and has become one of the important tools for nonlinear modeling.It is widely used in bioinformatics,medicine,social sciences and other fields,and has great potential in risk identification and early warning.Based on this paper,a fraud risk identification model of listed companies based on random forest is constructed.The related data processing and model architecture are programmed in Python environment.This paper firstly combs the papers from the exploration of the motivation of the fraud and the theoretical analysis,the fraud risk identification index and the fraud risk identification method,and discusses in detail the research results,frontier dynamics and existing deficiencies of the fraud risk identification model.On this basis,the random forest algorithm is introduced into the fraud risk research of listed companies,and introduces the basic principles,advantages and disadvantages of random forest and the feasibility of model construction.Secondly,select 430 A-share listed companies and the same number of matching companies that were fraudulent between 2014 and 2017 as research samples,and select indicators from the perspectives of corporate governance,information disclosure,special issues,industry pressure and financial stability.The system uses exploratory data analysis(EDA)and random forest algorithm to deeply mine the characteristics of primary selection indicators and repair index system,and then constructs the fraud risk identification model of listed companies based on random forest,and carries out model discrimination accuracy test and model performance comparison.The results show that the fraud risk model constructed in this paper has a good recognition effect,and has good performance and stable performance on high-dimensional samples.Finally,based on the above theoretical analysis and empirical research,the paper summarizes the full text and puts forward the inadequacies and research prospects of this paper.