Research On Enterprise E-Commerce Credit Risk Early-warning Based On Bagging

E-commerce,a new mode of business,is gaining its popularity during the boom in IT,within which some basic problems got their solutions step by step such as transactions and logistics.However problems remain to be solved as deception and credit risks happen in E-commerce activities,which became a bottleneck that impedes E-commerce from improving healthily.Based on the situation that rapid development came along with credit risk,the theory of this thesis offers an optimization for Enterprise E-Commerce Credit Risk Early-warning System,by the help of which early-warnings can be issued timely and accurately,in order that loses of enterprises are reduced,that relative departments of credit management are well-informed for the supervision of enterprises.The system is useful for both sides in the E-commerce trades,for the assessment of credit of buyers,sellers and competitors when further decisions are made.The thesis focuses on the information collecting and identification of risks.As for the information collecting of risks,the thesis made an examination on the index systems of the former theses via which two indexes-Online transaction dispute rate,R&D investment account for business income-are introduced with innovation,for a better reflex of service capacity and evolution capacity,while preserving part of the financial indexes to reach out for a better reflex of enterprise E-commerce credit risks.On the identification of risks,this thesis made an analysis that if SVM of the machine learning industry and integrated learning is feasible within the thesis.A notable performance of classification was reached under this innovation that integrated learning is introduced into the enterprise E-commerce credit early-warning industry and that the bagging algorithm is brought in for the optimization of SVM.It is seen that the accuracy of the identification of risk is higher after the optimization of the Enterprise E-commerce Credit Risk Early-Warning System,which speaks for the feasibility of the reasonably introduction of new index systems and integrated learning algorithm in this industry.The optimized early-warning system proposed in the thesis is practical due to the ability of self-learning of machine learning method.Furthermore,there are three degrees designed for the indication of warning,simply&clear,so that the early-warning system integrated in this thesis will be easier to be promoted.