Project Details

This project involves the use of artificial neural networks to predict NFL offensive plays. The focus was on creating models that can accurately learn and make predictions based on play-by-play data since 1999. The project explores the use of Long Short-Term Memory (LSTM) networks tailored to handle the sequential nature of football plays.

The project includes the following key features:

• Data preprocessing and normalization
• Model training and validation
• Hyperparameter tuning
• Model Development and Training
• Model Evaluation and Results

Achieved an 69.5% accuracy for specficially Patriots' offensive plays and 73% accuracy when expanding to all teams.

For more details, you can read my Medium article or check out the code on my GitHub.