hafiz-asad-2554/E-Commerce-Fraud-Detection-Prevention

# E-Commerce Fraud Detection & Prevention System **Storefront:** Secure Shopping **Submitted by:** Hafiz Muhammad Asad Mustafa (S23NDOCS1M01042) & Muhammad Islam (S22NDOCS1M01093) **University:** The Islamia University of Bahawalpur, Bahawalnagar Campus **Currency:** PKR Full-stack e-commerce with **rule-based fraud checks (Laravel)** and an **ML fraud engine (FastAPI)**. ## Project structure E-Commerce Fraud Detection & Prevention/ ├── ecommerce Site/ │ ├── frontend/ # HTML/CSS/JS storefront (port 5500) │ ├── backend/ # Laravel 12 API + admin (port 8000) │ ├── start-backend.bat │ └── start-frontend.bat └── fraud detection engine/ ├── backend/ # FastAPI + ML (port 8001) ├── datasets/ # Training CSV samples ├── model/ # Trained model artifacts ├── database/ # Sample transactions DB scripts ├── setup-ml-and-db.bat └── start-engine.bat ## Prerequisites - PHP 8.2+ and [Composer](https://getcomposer.org/) - Python 3.11+ - Git ## One-time setup ### 1. Laravel backend cd "ecommerce Site\backend" composer install copy .env.example .env php artisan key:generate php artisan migrate --seed php artisan storage:link Optional — hero slides and Kaggle-style product images: php artisan db:seed --class=HeroSlideSeeder php artisan db:seed --class=KaggleProductSeeder php artisan products:sync-local-images --max=4 php artisan db:seed --class=ContactSeeder Product images are read from `ecommerce Site\frontend\assets\images\products\` (CSV + `e-commerce/images/`). ### 2. Fraud detection engine cd "fraud detection engine\backend" python -m venv venv .\venv\Scripts\activate pip install -r requirements.txt cd .. .\setup-ml-and-db.bat This trains the model (`model/fraud_model.joblib`) and creates the sample SQLite DB (`database/fraud_transactions.db`). ## Run (3 terminals) | # | Command | URL | |---|---------|-----| | 1 | `ecommerce Site\start-backend.bat` | http://127.0.0.1:8000 | | 2 | `fraud detection engine\start-engine.bat` | http://127.0.0.1:8001/docs | | 3 | `ecommerce Site\start-frontend.bat` | http://127.0.0.1:5500 | | Page | URL | |------|-----| | Store | http://127.0.0.1:5500/index.html | | Checkout | http://127.0.0.1:5500/checkout.html | | Admin | http://127.0.0.1:8000/admin/login | | ML dashboard | http://127.0.0.1:8001/dashboard | **Admin login:** `admin@iub.edu.pk` / `password` Set `API_BASE_URL` in `ecommerce Site/frontend/js/config.js` to `http://127.0.0.1:8000/api` if needed. ## Testing ### FastAPI health & prediction curl.exe http://127.0.0.1:8001/health curl.exe -X POST http://127.0.0.1:8001/predict -H "Content-Type: application/json" -d "{\"amount_pkr\":200000,\"city\":\"Lahore\",\"ip_region\":\"sindh\",\"payment_method\":\"card\",\"previous_orders_24h\":5,\"account_age_days\":0.3,\"is_new_account\":true,\"items_count\":1}" Expect high `combined_score` / `"risk_level": "High"` for the sample above. ### Sample transactions database cd "fraud detection engine\database" python ..\database\seed_database.py .\view-sample-db.bat Or open `fraud_transactions.db` in [DB Browser for SQLite](https://sqlitebrowser.org/). ### Storefront fraud demos (Laravel rules) 1. **High amount** — laptop Rs. 145,000+ 2. **Location mismatch** — city Lahore vs simulated region Karachi 3. **Velocity** — 3+ orders in 30 minutes as the same user Review flagged orders in **Admin → Fraud Detection**. ## Features - Customer storefront: products, cart, multi-step checkout, geo hints, payment validation - Admin: products (multi-image upload), categories, orders, users, hero slides, fraud module - FastAPI: hybrid ML + rules, `/predict`, `/dashboard` with prediction logging ## API (Laravel) Base URL: `http://127.0.0.1:8000/api` | Method | Endpoint | Description | |--------|----------|-------------| | POST | `/register`, `/login` | Auth (Sanctum) | | GET | `/products`, `/categories`, `/slides` | Catalog & homepage | | GET | `/checkout/geo` | IP geolocation hint | | POST | `/orders` | Place order (fraud checks) | Admin panel routes are under `/admin/*` (session auth). ## Retrain ML model cd "fraud detection engine\backend" .\venv\Scripts\activate python train.py python ..\database\seed_database.py ## Troubleshooting | Problem | Fix | |---------|-----| | Login `Failed to fetch` | Start `start-backend.bat`; free port 8000 | | FastAPI `model not trained` | Run `setup-ml-and-db.bat` | | Port 8001 in use | Stop other Python/uvicorn processes | | Product images missing | `php artisan storage:link` and run image seeders | | Old cached checkout | Hard refresh (`Ctrl+F5`) on checkout page | ## Contact **Hafiz Muhammad Asad Mustafa** — hafizasad2554@gmail.com — S23NDOCS1M01042 **Muhammad Islam** — S22NDOCS1M01093 — Haroonabad, Pakistan — 0301 8445656 *Final Year Project — Department of Computer Science — IUB Bahawalnagar*

标签：ffuf