Study traditional and modern approaches used by banks to assess borrower creditworthiness.
Develop or evaluate models that predict the probability of loan repayment default.
Analyze which customer variables (e.g., income, credit history, DTI ratio) significantly influence credit risk.
Use algorithms such as logistic regression, decision trees, or neural networks to enhance prediction accuracy.
Provide actionable insights to help banks reduce NPA (non-performing assets) and improve loan approval processes.
Source historical banking/lending data and clean it for modeling (handle missing values, outliers, etc.).
Choose relevant variables and create new ones (e.g., credit utilization rate) to improve model performance.
Train classification models (e.g., logistic regression, random forest) to classify applicants into low/high risk.
Use metrics like accuracy, ROC-AUC, precision, and recall to validate and compare model effectiveness.
Prepare a detailed report explaining the methodology, results, and business recommendations, followed by a team presentation.
