Enhancing Credit Rating Precision for Financial institution Through Data Mining and Analytics

B.Sc. Badhan Saha

Master's thesis

Abstract:

This Thesis work explores how to improve credit scores in banks. It uses high level data mining and analytics to make credit score predictions more exact and trustworthy. The research uses logistic regression, CHAID, and CART — three well known models. It tests these models on a dataset full of different credit score situations. But training the models isn't the only step. This study also tests the models using historical data. Then it checks them against new, unseen data. This is like what happens in real life, with unpredictable future customers. The study doesn't just look at how the models do during training. It also looks at how they could work in real life banking. Lastly, the research looks at metrics like AUC and Gini. These show the power and accuracy of the model predictions. The models are put to the test with different credit score cases. The research looks at how well the models can tell good credit scores from bad ones. And it looks in detail at each model's strengths and weaknesses. This could help banks make strategic decisions. This study wraps up with tips for future work, highlighting the chance for more research in building and refining models. The idea is to overcome limits set by computer resources. It blends data mining and analytics perfectly. The goal? To boost the finance field with better accuracy and prediction in credit score checks.

Keywords

Credit Rating Enhancement, Data Mining, Logistic Regression, CHAID Model, CART Model, Evaluation Metrics, Financial Institutions, Data Analytics, Predictive Analytics

Language used: English

Date on which the thesis was submitted / produced: 26. 3. 2024

Thesis defence

• Supervisor: doc. Ing. Jan Tyrychtr, Ph.D.
• Reader: Richard Vágner, externi

Czech University of Life Sciences Prague

Faculty of Economics and Management

Master programme:
Informatics