A Significant Estimation and Prediction of High and Low Risk in Approving Bank Loan by Using Novel Creditworthiness by Comparing Support Vector Machine Model Over Logistic Regression
摘要
This study introduces a predictive model for assessing the risk associated with bank loan approvals, utilizing a novel metric of creditworthiness derived from a dataset of 500 customers, each evaluated across 10 distinct attributes. Employing two advanced classification methods—Support Vector Machine (SVM) and Logistic Regression—we conducted analyses at a stringent significance threshold of 0.05% and a robust statistical power of 80%. Our results, with a confidence interval set at 95%, demonstrated that Logistic Regression, boasting a 73.98% prediction accuracy, outperformed SVM's 65.04% in determining loan approval risks. This superiority indicates Logistic Regression as a more reliable method for credit risk assessment in the banking sector.