Research On Supply Chain Finance Credit Risk Evaluation Model Based On PCA-LSFSVM

Supply Chain Finance (SCF) service originated in the1990s, some domestic and foreignadvanced logistics companies and commercial banks launched a variety of supply chainfinance services to help companies manage their cash flow. In recent years, supply chainfinance has become the focus of many logistics companies and financial institutions. In China,the design of supply chain finance business is based on the structure, trading and operationalcharacteristics of supply chain, concentrated on the largest enterprises, considering severalSmall and Medium Enterprises (SMEs) involved in the supply chain. Taking the overall creditconditions of the supply chain in to consideration, it can provide a better financial service forSMEs, thereby promote the growth of the entire supply chain value.In the past credit business,the individual SMEs themselves’ were the mainly focus ofcommercial band credit risk evaluation, by analyzing their financial data, mortgage and othersubstances, commercial bank decided to provide loans or not. Supply chain finance model hasprovided a new and creative perspective to identify and evaluate the credit risks of SMEs. Inthe supply chain financing mode, the core enterprise’ financial conditions and the stability ofthe entire supply chain are also important reference factors in credit risk evaluation.In this paper, we studied how to evaluate SMEs’ credit risk in the perspective of supplychain finance. Firstly, by analyzing the characteristics of supply chain finance model, we havebuilt the supply chain finance credit risk evaluation index system. Then we use the PrincipalComponent Analysis (PCA) method to reduce the dimension of the index data, simplify thedata structure to improve classification efficiency. Secondly, Comparing the advantages anddisadvantages of existing credit risk evaluation methods, we choose the Support VectorMachine (SVM) as the credit risk evaluation method, and then we also introduce the fuzzymembership degree and variable penalty factor to improve SVM’s classification algorithm, onthe basis of all these, we construct the supply chain finance credit risk evaluation model.To simplify the calculation process, we translated complex quadratic programmingproblem into solving linear equations by introducing the quadratic loss function, and find outthe optimal classification function. Finally, through empirical research, we confirmed that thesupply chain finance credit risk evaluation model based on principal component analysis andfuzzy least squares support vector machine (PCA-LSFSVM) is better than the supply chainfinance credit risk evaluation model based on binary Logistic regression. It can solve theproblem of supply chain finance credit evaluation effectively, thus this paper provides a scientific and rational tool for commercial bank in supply chain finance credit risk evaluation,thus it has important theoretical and practical significance.