This project aims to predict traffic speed using time-series forecasting techniques. By leveraging XGBoost, the model forecasts future speeds based on historical sensor data. Accurate traffic speed prediction helps in traffic management, route optimization, and smart city planning.

• Dataset: METR-LA (Traffic speed sensor data in Los Angeles)
• Type: Time-series dataset with multiple sensors
• Features Engineered:
  • Timestamp-based Features: Hour, Day, Weekday, etc.
  • Rolling Statistics: Moving averages for trend analysis
  • Lag Features: Past values to predict future speeds

Predict the next time step's traffic speed for multiple sensors using ML models like XGBoost.

✅ High accuracy with minimal errors makes this model useful for real-world applications.

• Programming Language: Python
• Machine Learning Model: XGBoost
• Libraries Used:
  • pandas, numpy (Data Processing)
  • xgboost, sklearn (Machine Learning)
  • matplotlib, seaborn (Visualization)

1. Install Dependencies

   pip install -r requirements.txt

2. Create required file

   python traffic_prediction.py

3. Load and use the Model

   import xgboost as xgb
   
   model = xgb.Booster()
   model.load_model("xgb_traffic_model.json")

• ✅ Multi-Sensor Forecasting – Predicts speed across multiple locations
• ✅ Feature Engineering – Uses lag features, rolling statistics, and time-based features
• ✅ Optimized ML Model – Fine-tuned XGBoost for best performance
• ✅ Scalable – Can be extended to include more real-world variables

• 📡 Integrate live traffic APIs for real-time forecasting
• 🗺️ Deploy as a web service for traffic management systems

This project is open-source under the MIT License.