Exploring impact on auto loan assessment and predictive modeling of credit scores with machine learning Algorithms

Abstract

The main goal of Credit Information Bureau India Limited (CIBIL) in the auto loan procedure is to give lenders vital information regarding borrowers’ creditworthiness. CIBIL scores help assess loan risk. This study uses advanced ML techniques, Gaussian Process Classification (GPC) and Gradient Boosting Classification (GBC), to predict scores, improving accuracy and reliability in evaluating creditworthiness for vehicle loan approvals. To enhance the accuracy of these predictive models, the Electric Eel Foraging Optimization (EEFO) and the Political Optimizer Algorithm (POA) are incorporated as optimization methods. By integrating these ML models with sophisticated optimization algorithms, a highly accurate prediction of CIBIL scores is aimed to be achieved. This can improve the efficiency and reliability of the auto loan approval process. GBEE excelled with top accuracy in both training (0.965) and testing (0.903) phases. GBPO closely followed, showing robust predictive power. GBC was reliable, particularly in high and mid-probability conditions, despite trailing GBEE and GBPO.