Emmanuel902-ai/aml-mobile-money-rwanda

Code and experiments for my MSc thesis on AML risk detection in Rwanda mobile money transactions. 💸 Detecting Money Laundering Risks in Rwanda Mobile Money Transactions A Machine Learning Approach for AML Risk Detection Nahimana Emmanuel 📧 nahimana.emmanuel@aims-senegal.org African Institute for Mathematical Sciences (AIMS) — Senegal Supervised by: Dr. Yaé Ulrich Gaba AI Research and Innovation Nexus for Africa (AIRINA) · AIRINA Labs by AI.Technipreneurs · Cotonou, Bénin 🎓 Master of Science in Data Science (Big Data) African Institute for Mathematical Sciences (AIMS) Senegal --- # 📑 Table of Contents 1. Project Overview 2. Abstract 3. Abbreviations 4. Introduction 5. Research Objectives 6. Research Questions 7. AML Methodology Pipeline 8. Models Implemented 9. Model Performance 10. Visual Results 11. Rwanda Context and Relevance 12. Conclusion and Future Work 13. Acknowledgements 14. Appendix: Code Implementation 15. Repository Structure --- # 📌 Project Overview This repository contains the **code, notebook, and supporting materials** for my MSc thesis titled: **“Detecting Money Laundering Risks in Financial Transactions in Rwanda: A Machine Learning Approach with a Case Study on Mobile Money.”** The research investigates how **machine learning techniques can support Anti-Money Laundering (AML) monitoring systems** by detecting suspicious patterns in large-scale financial transaction data. Mobile money has become a cornerstone of financial inclusion in Rwanda and across Sub-Saharan Africa. However, the rapid growth of digital transactions also increases the risk of **money laundering and illicit financial flows**. This project develops a **data-driven AML detection pipeline**, combining: - Data preprocessing - Feature engineering - Supervised classification - Anomaly detection - Hybrid model fusion - Explainability and risk scoring The objective is to explore how machine learning models can enhance **transaction monitoring systems used by financial institutions and regulators**. --- # 📋 Abstract In this thesis we develop a mach …