A full-stack system that predicts product return probability using machine learning and intelligently routes returned items (restock, refurbish, inspect) using business logic. Built for e-commerce platforms to optimize reverse logistics and reduce operational losses.
E-commerce companies suffer heavy losses from returned products due to:
- High return rates during sale seasons
- Inefficient handling of returns (restock vs refurbish vs recycle)
- Manual decisions costing time and money
🧠 This project solves it by predicting returns before they happen and suggesting smart routing decisions for returned products.
- 🔮 Predicts return probability using Random Forest
- 🚦 Smart routing logic based on return reason and product type
- 🧾 Admin dashboard to view total orders and returns
- 🛍️ User interface for placing orders
- 🔗 Backend APIs for ML inference and data management
| Layer | Tech |
|---|---|
| Frontend | React.js |
| Backend | Flask / FastAPI (Python) |
| ML Model | Scikit-learn + joblib |
| Database | PostgreSQL |
| Deployment | Streamlit Cloud / Render |