mfyumsak/NFL_EPA_Prediction_App

🏈 NFL EPA Prediction App

This Streamlit application allows users to predict Expected Points Added (EPA) for NFL plays using pre-trained machine learning models based on tracking and play-level data from the NFL Big Data Bowl 2021 competition.

🚀 What We Offer

• —📂 Upload your own custom play-level data
• —🤖 Predict EPA using a pre-trained RandomForestRegressor model
• —📈 Download the results as CSV for further analysis
• —🎓 Use provided sample CSVs to test the app without needing any dataset

🧠 Model Information

• —Model Type: RandomForestRegressor
• —Evaluation Results:
• —RMSE: 0.2293
• —R² Score: 0.9824
• —Trained on:
• —Features like ball_carrier_speed, acceleration, direction, and one-hot encoded ball_carrier_position

📁 Project Structure

├── app.py ├── model.pkl ├── sampleplaysfast.csv ├── sampleplaysmedium.csv └── sampleplaysslow.csv