SCORING METHODS IN CREDIT RISK

SCORING METHODS IN CREDIT RISK

Abhilash Satheesh, Mrs. Shruthi K S, Murali Krishna V Hegde, Ram Maruti Metre, Prerith S K

Abstract— Credit scoring plays a crucial role in the financial industry, enabling lenders to assess the creditworthiness of potential borrowers. This project aims to implement and evaluate various credit scoring methods using a real-world dataset from Kaggle. The dataset contains information about loan applicants, including demographic data and credit history. To prepare the data for analysis, preprocessing steps such as handling missing values and removing outliers were performed. Five credit scoring methods were implemented using Python and scikit-learn: K- Nearest Neighbor (KNN), Logistic Regression, Decision Tree, Random Forest, and Support Vector Machine (SVM). The dataset was split into training and testing sets, and each model was evaluated using accuracy, sensitivity, and specificity metrics. The results demonstrated the effectiveness of the implemented models in predicting credit risk, with KNN, Logistic Regression, and SVM exhibiting the highest accuracy scores. Decision Tree and Random Forest also showed promising performance. This project highlights the importance of implementing and evaluating multiple credit scoring methods to identify the most suitable approach for a given dataset. The findings can assist financial institutions in making informed lending decisions and managing credit risk effectively. Future work could explore additional features, ensemble methods, or advanced techniques to further enhance the predictive power of credit scoring models.

Keywords— Credit scoring, Machine learning, Risk assessment, Predictive modeling, Financial lending.