K-BEE1709/fraud-detection-mobile-money

Decision Tree fraud detection for mobile money transactions with Ghana cedis support # 🇬🇭 Mobile Money Fraud Detection System A machine learning system that detects fraudulent mobile money transactions using a Decision Tree classifier. Built with Ghana Cedis (₵) support for local relevance. ## 🎯 Key Results - *82.2%* fraud detection rate (recall) - *277× better than random* (PR-AUC 0.36 vs 0.0013) - Real-time transaction scoring ## 📊 Dataset - *Source:* PaySim synthetic dataset (Kaggle) - *Size:* 6.36M transactions - *Fraud rate:* 0.13% (highly imbalanced) ## 🛠️ Tech Stack - Python 3.11 - scikit-learn (Decision Tree) - pandas, numpy - matplotlib, seaborn ## 🚀 Quick Start ### Installation ```bash git clone github.com cd fraud-detection-mobile-money pip install -r requirements.txt