Loan Charge-Off Forecasting Analytics

The objective of this project was to develop a predictive analytics solution to forecast loan charge-offs for a credit union. By analyzing historical loan performance, member financial behavior, and external economic indicators, the model enabled early identification of high-risk loans and supported proactive risk mitigation strategies.

The solution helped the credit union move from reactive charge-off management to a data-driven, preventive approach, improving overall portfolio health and financial stability.

Collected and analyzed historical loan data, including repayment behavior, delinquency records, and member credit profiles

Built a machine learning model to predict the probability of a loan becoming a charge-off

Developed an interactive Power BI dashboard to track charge-off trends and highlight high-risk loan segments

Extracted loan-level data such as outstanding balances, payment delays, interest rates, and credit utilization

Engineered key risk indicators including Debt-to-Income (DTI) ratio, payment behavior trends, and delinquency severity scores

Cleaned and prepared datasets to ensure modeling accuracy and consistency

Trained classification models such as Logistic Regression, Random Forest, and XGBoost to predict charge-off risk

Evaluated model performance using metrics including precision, recall, and ROC-AUC to ensure reliable forecasting

Selected the most effective model based on accuracy, interpretability, and business usability

Designed a Power BI dashboard to visualize:

Delivered actionable insights to support early interventions, such as targeted outreach and repayment restructuring

Early Risk Identification: Enabled the credit union to flag high-risk loans well before default

Improved Loan Management: Supported smarter lending policies and risk-based pricing decisions

Reduced Financial Losses: Helped minimize charge-offs through proactive engagement and preventive strategies

This project strengthened the credit union’s credit risk management framework, improved visibility into loan portfolio health, and enabled data-driven decision-making to reduce financial losses and enhance long-term stability.