🏁 F1 Pit Stop Predictor

A robust, machine learning-powered web application that predicts whether a Formula 1 driver will pit on the next lap based on live race telemetry.

📸 Screenshots

🚀 Features

• Advanced ML Pipeline: Built using an XGBoost Classifier engineered specifically for time-series race data.
• Data Leakage Prevention: Trained using GroupShuffleSplit on race events to ensure the model generalizes perfectly to entirely new and unseen races.
• Handling Class Imbalance: Predicts rare pit stop events with high recall (85%) by explicitly penalizing false negatives via scale_pos_weight.
• Rolling Time-Series Features: Uses rolling lap time averages and lap time gradients to evaluate a driver's true degradation curve.
• Sleek Web Interface: Features a premium, dark-mode, F1-styled Streamlit application.

🛠️ Installation & Usage

1. Clone the repository.
2. Install the requirements:

   pip install -r requirements.txt

3. Run the Streamlit web application:

   streamlit run app.py

🧠 Model Architecture

• Algorithm: XGBoost
• Feature Engineering: Pandas grouping (3-lap rolling averages, LapTime gradients).
• Hyperparameter Optimization: Randomized Grid Search (n_estimators, max_depth, learning_rate).