Mathematics
Sports Prediction
Sports Prediction Using Neural Networks
Student Researchers: Andrew Blaikie and Gabriel Abud
Faculty Advisors: John David and Drew Pasteur (Mathematics)
This project was a continuation of a 2010 AMRE/HHMI project dealing with artificial neural networks (ANNs) as a tool for predicting NFL football games. We built multiple ANN models to predict both college and professional football games. ANNs have many uses in today's scientific fields and are an efficient way to model complicated systems. We devised our most efficient model by analyzing several years of game statistics, using methods including correlation, principal component, derivative-based, and linear regression analysis. Predicting college football was a more difficult problem due to the wide variety of team abilities and schedule strengths, so the results were not as accurate as they were for the NFL model. Additionally, we collected large amounts of data on other sports for future research.