Research Of Commercial Bank Credit Risk Assessment Based On The Method Of Dynamic Fuzzy Integration Support Vector Machine

Currently,Commercial bank credit risk has become one of the most con centrated expression formstate of our country’s main financial risk, it has the extremely important influence to the whole economic operation.In the study of this article reviewed several kinds of elements that may affect the risk,and every kinds of specific information.Finally on the basis of systematic, comprehensive, scientific, accessibility principles of s electing the corresponding source index and build into a system.And are n ot sources of each individual isolation,but considered systemic associations, build into a whole to study.This article will ultimately selected 16 indicato rs as members of the system,through statistical analysis points to four fact ors have meaning they belong to.The system more comprehensive reflect t o predict the likelihood of enterprise credibility and can pay in time.The core of this article is for the research of prediction models.We wi 11 review the development of credit risk assessment method,and various sta ges of scholars adopt method principle and the advantages and disadvantag es as well as perfecting the process of improvement.According to the the drawbacks of the old method, this paper puts forward a new improved met hod——The integration of fuzzy dynamic support vector machine (SVM) model.In this paper, we select SVM as the foundation algorithm to suppo rt the model.Considering the reality,identify the bad side is always is more important than a good party.In addition, in reality, an individual is not ab solutely belongs to a category, on the contrary is not entirely do not belo ng to another category.Introduces the punishment of variable parameters an d the corresponding fuzzy membership degree.This paper uses the fuzzy int egral integration strategy.Finally comprehensive utilization of the SVM and fuzzy integral and fuzzy membership degree, variable penalty factor, the 1 east squares principle.We will eventually programming model, and compared with other mode Is, and finally confirmed the improved method is feasible and high accurac y, the research into the meaning of theory and practice.