| dc.description.abstract | This thesis first presents a least squares regression model to identify the in-game statistics that help explain the variation in point spread for NCAA Division II Women’s Basketball Tournament games. Then a logistic regression model is presented to estimate the probability of a team winning a tournament game based on the differences in significant in-game statistics. Differences in the following variables are significant in both models: field goal percentage, 3-point field goal percentage, free throw percentage, offensive rebounds, personal fouls and turnovers. Difference in assists is only significant in the point spread model. Both models are validated using the in-game statistics for the 2015 tournament, indicating a prediction accuracy as high as 95.24%. Seasonal averages for the 2014 - 2015 season are then used to predict game results in the 2015 tournament. The prediction accuracies are 60.32% and 66.67% for the point spread model and the logistic regression model, respectively. | en_US |
