Enhancing Credit Risk Prediction in Financial Services Using Logistic Regression and Artificial Intelligence Techniques

Abstract

The advent of artificial intelligence (AI) and big data has transformed the financial industry, particularly in credit risk analysis and prediction. Traditional manual methods of assessing credit risk are no longer feasible for managing the increasing complexity of data and growing customer base in modern banking. This study examines the application of logistic regression and other AI-driven models, such as K-nearest neighbors (KNN), in predicting credit defaults. Findings demonstrate that logistic regression surpasses KNN in accuracy by 15.49%, showing greater efficiency in processing large datasets for credit assessment. Additionally, the performance metrics—ROC (receiver operating characteristic) curve and AUC (Area Under the ROC Curve)—confirm logistic regression’s robustness, making it a practical solution for real-world applications in financial risk management. While logistic regression is advantageous for binary classification tasks within the financial sector, limitations include its reliance on data quality and applicability within varied banking environments.