Predicting car loan approval probability using credit information reports through a machine learning approach

This study examines how credit information reports (CIRs) predict car loan approval probabilities. Accurate prediction models are essential for financial institutions to mitigate risks and make informed lending decisions. CIRs arrive at judgments about creditworthiness based on data construction of credit history. Machine learning techniques using advanced methods enhance predictive capacity by splitting large data bases to determine trends beyond the scope of human testers. Two grand models, bagging classification and extreme learning machine, are employed in this case. These models are subsequently optimized using two methods, probabilistic uncertainty-aware multi-objective algorithm optimization and Fox optimization (FO), to create four hybrid models: ELFO, ELPO, BAFO, and BAPO. The outcomes reveal that the hybrid models outperform their versions. Specifically, the BAPO model was ranked highest with overall accuracy of 0.944, training accuracy of 0.963, and testing accuracy of 0.900, together with improved precision, recall, and F1-scores. The result demonstrates the ability of optimization techniques to enhance the predictive performance of the models, which can be useful input for banks to decide the likelihood of lending. With the application of machine learning, the study provides a solid foundation for enhancing financial industry decision-making processes.