Enhancing Credit Note Precision and Competitive Positioning in B2B Transactions Through Data Analytics

This thesis delves into the critical realm of credit rating enhancement within financial institutions, with a primary focus on the dynamic landscape of banks. Employing advanced techniques in data mining and analytics, the research endeavors to revolutionize the precision and reliability of credit rating predictions. The study meticulously employs three powerful models—logistic regression, CHAID, and CART—leveraging a comprehensive dataset that captures diverse credit scenarios. Beyond ordinary model training, the research introduces a groundbreaking approach. The models are not only rigorously trained on historical data but are also subjected to rigorous testing on previously unseen data, simulating real-world scenarios and future client unpredictability. This dual-phase assessment not only provides a thorough analysis of model agreement during training and testing but also underscores the practical applicability of the proposed methodologies in the actual practices of banking institutions.