How hamida-paiman/fraud-detection-system-using-ensemble-machine-learning-models works
This is an end-to-end e-commerce transaction fraud detection project: transaction data is prepared and engineered, supervised models are compared, and a trained pipeline supports interactive predictions. The primary user workflow is entering transaction details in the Flask app, which derives features, scores the transaction, and displays a verdict and probability. A separate command-line assistant also collects transactions and prints a prediction with rule-based risk factors. The README documents modeling and dataset work, but notebook internals were not sampled; the supplied model-loading paths also differ from the tree.
Main entry points: web_app.py Flask prediction · fraud_chatbot.py Chat interaction · fraud_model.pkl Trained pipeline · final_submission_1.ipynb Model experiments
Components 14
Data and modeling
- Dataset preparation
- Model experiments
final_submission_1.ipynb - Trained pipeline
fraud_model.pkl
Web predictions
- Transaction form
index.html - Flask prediction
web_app.py - Feature engineering
web_app.py - Prediction result
index.html
CLI assistant
- Chat interaction
fraud_chatbot.py - Feature engineering
fraud_chatbot.py - Risk explanation
fraud_chatbot.py
Other components
- Data scientist
- Fraud datasets
- Web user
- CLI user
Connections 14
- Fraud datasets → Dataset preparation: provides data
- Data scientist → Dataset preparation: prepares data
- Dataset preparation → Model experiments: prepares features
- Model experiments → Trained pipeline: trains pipeline
- Web user → Transaction form: enters details
- Transaction form → Flask prediction: submits form
- Flask prediction → Feature engineering: builds features
- Flask prediction → Trained pipeline: scores transaction
- Flask prediction → Prediction result: renders result
- CLI user → Chat interaction: provides details
- Chat interaction → Feature engineering: builds transaction
- Chat interaction → Trained pipeline: predicts transaction
- Chat interaction → Risk explanation: requests explanation
- Feature engineering → Risk explanation: supplies factors