Data analysis and machine learning project to predict and reduce loan defaults in Ghana's lending sector
# Loan-Default-Analysis-Ghanaloanconnect
Data analysis and machine learning project to predict and reduce loan defaults in Ghana's lending sector
# Loan Default Analysis: Ghana Lending Sector
## 📋 Project Overview
This data analytics project addresses the critical challenge of loan defaults in Ghana's non-bank financial institution (NBFI) sector. GhanaLoanConnect, a peer-to-peer lending platform, faces a 16.01% default rate—significantly above industry standards—threatening its sustainability and financial inclusion mission.
## 🎯 Business Objectives
1. **Identify High-Risk Segments**: Pinpoint sectors and borrower profiles with the highest default probability
2. **Diagnose Root Causes**: Analyze factors driving repayment difficulties
3. **Develop Mitigation Strategies**: Recommend data-driven interventions to reduce defaults by 20%
## 💡 Key Insights & Impact
Analysis of 9,578 loan records revealed critical risk patterns:
* **Small Business loans** showed the highest vulnerability with a **28% default rate**, followed by Educational loans at **20%**
* **Credit scores** demonstrated strong predictive power - borrowers with FICO 600-649 had a **32% default rate** vs **7%** for 750+
* **Interest rate correlation** was evident - loans >15% interest had **24.8% default rate** vs **4.2%** for rates <8%
* **Machine learning model** achieved **84% accuracy** in predicting high-risk borrowers
## 🚀 Recommendations & Implementation
A phased strategy was developed to transform risk management:
1. **Risk-Based Pricing:** Implement tiered interest rates (8%-18%) based on borrower risk profiles
2. **Sector-Specific Underwriting:** Apply enhanced due diligence for Small Business and Educational loans
3. **Automated Decision Systems:** Integrate ML model to flag high-risk applicants and reduce processing time by 65%
4. **Portfolio Diversification:** Rebalance toward lower-risk segments (Major Purchases: 11% default, Credit Cards: 12%)
## 📈 Expected Business …