ML-powered assistant coach for football play-calling. Enter venue, matchup, and game situation to get top play recommendations with success probability.

You need two terminals: one for the backend API, one for the frontend. Run both from the repo root.

• Python 3.10+ (for backend)
• Node.js 18+ and npm (for frontend)

From the repo root:

# Create and use a virtual environment (recommended)
python3 -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r backend/requirements.txt

# Start the API server
uvicorn backend.main:app --reload

Backend runs at http://127.0.0.1:8000.

Model files: The backend loads .pkl files from the models/ folder. If you pulled the repo and that folder contains the trained artifacts, you’re set. If not, train the model first (see Training below).

In a second terminal, from the repo root:

cd frontend
npm install
npm run dev

Frontend runs at http://localhost:5173.

1. Open http://localhost:5173 in your browser.
2. Setup: Enter city (and click Fetch for weather), surface, roof, your team name, and opponent. Fill in your QB/RB/WR stats, then click Start game.
3. Game: Choose Offense or Defense. On offense, enter drive stats (scores, timeouts, starting yard line) and save. Use the clock (Play / Stop); when you press Stop on offense, enter yards gained for the play. Use Get top 3 plays to see recommendations.
4. Use ← Setup to return to the setup screen (you’ll be asked to confirm so you don’t quit the game by mistake).

The frontend proxies /api requests to the backend, so no CORS or port configuration is needed as long as both servers are running.

If the model artifacts are missing or you want to retrain:

1. Place your training data as data/train_with_stats.csv (or run the pipeline: merger.py → filter.py → merged2.py to build it from raw data).

2. From the repo root:

   python scripts/train_model.py

3. This writes .pkl files into models/. Restart the backend if it’s already running.