mwandirakings-prog/malawi-banking-analytics
Domain:
socioeconomicRecord type:
software
Creator:
mwa
Host:
Financial Performance and Credit Risk Analytics Dashboard for Malawi Banking Sector
# Malawi Banking Analytics Dashboard
### Financial Performance & Credit Risk Analytics Platform
---
## Project Overview
A fully deployable, enterprise-grade analytics platform that monitors
the financial performance of Malawi's major commercial banks and
assesses credit risk using advanced machine learning models.
This project is built to international standards including:
- Basel III/IV capital adequacy framework
- IFRS 9 Expected Credit Loss (ECL) modeling
- SHAP model explainability (EU GDPR compliant)
- Production-grade cloud pipeline architecture
---
## Banks Covered
| # | Bank |
|---|------|
| 1 | National Bank of Malawi |
| 2 | Standard Bank Malawi |
| 3 | First Capital Bank |
| 4 | NBS Bank |
| 5 | FDH Bank |
| 6 | Ecobank Malawi |
**Period:** 2018 — 2023
---
## Technology Stack
| Layer | Tools |
|-------|-------|
| Data Collection | Python, pdfplumber, pandas |
| Machine Learning | XGBoost, Scikit-learn, SHAP |
| Statistical Modeling | R, ggplot2, caret |
| Visualization | Power BI, matplotlib, seaborn |
| Cloud Pipeline | AWS Lambda, Apache Airflow |
| Data Warehouse | Google BigQuery |
| Engineering | GitHub, Docker, pytest |
---
## Model Performance
| Metric | Score |
|--------|-------|
| AUC-ROC | **0.8161** |
| Gini Score | **0.6323** |
| CV Mean AUC | **0.8184** |
| Training Records | 100,000 |
---
## Project Structure
malawi-banking-analytics/
├── data/
│ ├── raw/ # Source data from RBM and banks
│ ├── processed/ # Cleaned data and model outputs
│ └── synthetic/ # 100,000 synthetic loan records
├── src/
│ ├── data_collection.py # CAMELS ratio calculation
│ ├── synthetic_loans.py # Loan data generator
│ └── credit_risk_model.py # XGBoost + SHAP model
├── notebooks/ # Jupyter EDA notebooks
├── tests/ # pytest unit tests
├── airflow/ # Pipeline DAG definitions
└ …