Senior Capstone

Enhancing College Football Bowl Game Prediction using Linear Regression

In college football, analysts are always developing methods to predict outcomes. The most successful methods include statistical models and machine learning techniques. Most models rely on historical data from previous games. However, a current problem is the sudden absence of players, due to these players having the option to opt out of end of season bowl games.  This leads to holes in statistical data used for prediction. Therefore, this paper analyzes and enhances a previous successful prediction model while taking into account the impact of player opt-outs. This method uses linear regression and data analytics to identify significant variables based on the absence of key players and point differentials. The significant variables are then added into the model to create an improved method to predict the outcome of college football bowl games. With this system in place, fans, analysts, and sports bettors can have a better understanding of how teams will perform in bowl games. The purpose of this proposed research is to create an enhanced prediction method that benefits the game of football from an analytical viewpoint.