A comparative analysis of data mining methods in predicting NCAA bowl outcomes

Predicting the outcome of a college football game is an interesting and challenging problem. Most of the previous studies concentrated on ranking the bowl-eligible teams according to their perceived strength, and using these rankings to predict the winner of a specific bowl game. In this study, using eight years of data along with three popular data mining techniques (i.e., artificial neural networks, decision trees and support vector machines) we developed both classification as well as regression type models to assess the predictive ability of different methodologies (classification versus regression-based classification). In the end, the results showed that the classification type models predict the game outcomes better than regression-based classification models, and among the three classification techniques, decision trees produced the best results with better than 85% prediction accuracy on the 10-fold cross validation sample. The sensitivity analysis on trained models revealed that non-conference team winning percentage and average margin of victory are the two most important variables among the 28 that were used in this study.