IMANE-HR/sovereign-credit-risk

ML early warning system for sovereign default risk in Africa # Sovereign Credit Risk Modeling for African Emerging Markets > **Predicting sovereign default probability using macroeconomic indicators — with a focus on the underrepresented African continent.** --- ## Motivation Sovereign credit risk — the probability that a government defaults on its debt obligations — is one of the most consequential problems in international finance. It affects bond pricing, foreign investment flows, IMF bailout decisions, and ultimately the livelihoods of millions of citizens. While the academic literature is rich with models for advanced economies and Latin America, **Sub-Saharan Africa and North Africa remain systematically understudied**. Rating agencies such as Moody's and S&P cover fewer than 30 African sovereigns, leaving a significant information gap for investors and policymakers alike. This project builds a machine learning pipeline that: 1. Assembles a panel dataset of macroeconomic indicators for **54 African countries** from 1995–2022 2. Trains and benchmarks multiple classifiers to predict sovereign distress episodes 3. Uses **SHAP explainability** to identify the most important leading indicators 4. Produces actionable insights for Ministry-level policy analysis and debt management offices --- ## Project Structure ``` sovereign-credit-risk/ │ ├── data/ │ ├── raw/ # Original downloaded datasets (never modified) │ │ ├── world_bank_macro.csv │ │ ├── imf_debt_data.csv │ │ └── default_events.csv │ └── processed/ # Cleaned, merged, feature-engineered data │ └── panel_final.csv │ ├── notebooks/ │ ├── 01_data_collection.ipynb # API calls, data assembly │ ├── 02_eda.ipynb # Exploratory data analysis │ ├── 03_feature_engineering.ipynb # Lag features, ratios, transformations │ ├── 04_modeling.ipynb # Model training and evaluation │ └── 05_shap_interpretation.ipynb # Explainability analysis │ ├── src/ │ ├── __init__.py │ ├── …