Optimization Strategy of Credit Scoring System based on Support Vector Machine

Abstract

This article proposes a novel optimization strategy for credit scoring systems that exploits the capabilities of SVM. Focusing on the importance of personal credit scoring in today's credit dynamics, the article explores SVM's versatility in various domains through a literature review. The theoretical background underscores the unique approach and computational efficiency of SVM. The optimization strategy encompasses four critical aspects: debt solvency, earning potential, operational prowess, and growth capability using metrics such as asset-liability ratios. Experimental validation with credit card datasets from Australia and Germany illustrates the nuanced relationship between different K-values and performance metrics, and demonstrates the adaptability of SVM in improving credit scoring. In short, the article presents an original, comprehensive approach to credit risk management that integrates theoretical foundations, literature findings, and empirical experiments to improve the accuracy of credit scoring in the dynamic economic landscaper.