annampinganasalom-art/namibian-credit-risk-engine

End-to-end retail credit risk modeling engine with Bank of Namibia stress testing, XGBoost scoring, SHAP explainability, and interactive Streamlit UI. # 🏦 Namibian Retail Credit Risk & Alternative Data Scoring Engine An end-to-end Machine Learning credit decisioning engine built specifically for retail lending in Namibia. The application combines traditional risk modeling with macro-prudential stress testing (Bank of Namibia repo rate variations) and SHAP explainability. ## 📌 Project Overview * **Data Processing & Engineering:** Adapted public credit data into Namibian Dollars ($\text{N\$}$) with custom feature engineering (DSTI ratios, employment stability, macro sensitivity flags). * **Machine Learning:** Logistic Regression (Regulatory Baseline) vs XGBoost Classifier (Challenger Model). * **Explainability:** SHAP (SHapley Additive exPlanations) waterfall plots for interpretable credit approvals/declines. * **Interactive UI:** Streamlit web app providing real-time macro-economic scenario testing. ## 🚀 How to Run Locally 1. **Clone the repository:** ```bash git clone cd namibian-credit-risk-engine