This project is an AI-driven predictive sanitation system designed for smart cities. It forecasts garbage overflow risk using machine learning and assigns a Cleanliness Risk Score to urban zones.
The system enables preventive waste management instead of reactive collection.
- Garbage overflow prediction using XGBoost
- Cleanliness Risk Score (0–100)
- Risk categorization (Low / Medium / High)
- FastAPI-based REST backend
- Modular ML architecture
- AMD-optimized AI computation
- Structured backend project design
- Python
- FastAPI
- XGBoost
- Scikit-learn
- Pandas & NumPy
- Joblib
- Uvicorn
- Collects urban sanitation indicators (population, rainfall, complaints).
- Applies feature engineering.
- Uses XGBoost to calculate risk probability.
- Converts probability into risk score (0–100).
- Categorizes into Low, Medium, or High risk.
pip install -r requirements.txt
python backend/model_training.py
python -m uvicorn backend.app:app
- Real-time IoT bin integration
- Heatmap visualization dashboard
- Time-series forecasting
- Cloud deployment
- Multi-city scaling
Model training and inference are optimized for AMD Ryzen processors and Radeon GPU acceleration to support efficient AI computation.