| The purpose of this study was to determine if a predictive model could be developed to determine if there are certain variables that would indicate which Montana University/College a Montana high school football player may attend.; One hundred and fifty three student-athletes (football players) at four Montana higher education units (University of Montana-Missoula, Montana State University-Bozeman, Montana Tech of the University of Montana, and Carroll College) completed the survey. The survey consisted of 50 questions, divided into three areas: athletics, academics, and demographics. An artificial neural network (ANN) was employed for data analysis.; Six different groups of independent variables (tests) were run. The numbers of independent variables used in each test ranged from 143 in test one to 36 in test five. Each test was run twice (Tests A and B), using different groups of data points for the training, testing, and validation data sets in the artificial neural network.; Two metrics were used to determine the efficacy of each test: an R-value greater than or equal to .70 and a predictive accuracy of 75%, or greater. At least one run in each test provided results above the threshold limits for each metric. Six months after the data set used to develop the artificial neural network was collected, 20 student-athletes participating in the football program at Montana Tech were surveyed. The data contained in these surveys were used to validate the artificial neural network models developed in the first six tests.; The predictive accuracies of all four schools were above the threshold limit using this validation data set. The R-values for three of the four schools were above the threshold limit.; The results from this study indicate that artificial neural network technology shows promise as a predictive tool for collegiate coaching staffs, allowing them to use their recruiting budgets more efficiently. |
