Internet-based supply chain financing-oriented risk assessment using BP neural network and SVM.

Area of Science:

• Financial Engineering
• Computational Finance
• Risk Management

Background:

• Internet-based Supply Chain Financing (SCF) presents unique credit risks.
• Existing credit risk evaluation (CRE) methods require optimization for SCF products.
• Effective risk management is crucial for the stability of SCF markets.

Purpose of the Study:

• To optimize and evaluate an Internet-based SCF-oriented Credit Risk Evaluation (CRE) method.
• To develop a robust model for predicting and mitigating credit risks in SCF.
• To provide actionable insights for commercial banks in China.

Main Methods:

• Established a Risk Assessment Index System (RAIS) by identifying 12 key SCF risk factors.
• Explored principles of Backpropagation (BP) Neural Network (NN), Support Vector Machines (SVM), and Genetic Algorithm (GA).
• Implemented and validated a CRE model using the BP-GA approach in MATLAB with SCF risk assessment samples.

Main Results:

• The BP-GA model demonstrated high prediction consistency with actual classifications.
• Achieved a classification accuracy of 97.19% for test samples in the Internet-based SCF-oriented CRE system.
• Reported Type I and Type II error rates of 7.2% and 14.21%, respectively, for the BP-GA based CRE system.

Conclusions:

• The proposed BP-GA model offers a suitable and effective SCF-oriented CRE method.
• The study provides scientific and feasible suggestions for managing SCF credit risks.
• This research contributes to more reasonable and effective credit risk management in China's commercial banks.