Explainable Machine Learning System for Credit Default Prediction in Indian Micro-Lending

Objective: To engineer a deployable and explainable machine learning platform that predicts credit default risk for micro-lender applicants using real borrower data.

Micro-lenders in India face significant risks due to limited borrower information, leading to high default rates that threaten both financial viability and borrower inclusion, particularly among low-income and rural populations.

The proposed solution is an end-to-end explainable ML system that ingests alternative and traditional data (demographics, transaction history, mobile usage, etc.), trains classification models, and outputs both default risk predictions and interpretable explanations for loan officers and applicants.

Key deliverables include a user-facing web dashboard for loan officers, an ML backend with integrated SHAP/LIME-based explainability, integration with sample microfinance datasets (e.g., UCI/Indian MFIs), model performance benchmarking, and a working demonstration with simulated applicant data.

This system enhances transparency, reduces financial risk, and enables responsible credit expansion for Indian micro-lenders, with scalability for wider fintech, rural banking, and regulatory compliance applications.