Identifying significant features for Player Evaluation in NFL comparing ANNs and Traditional Models

The evaluation of player performance in sports is popular and important in modern sports, enabling teams to use real data in the construction of their rosters. This dissertation proposes to apply machine learning algorithms to predicting the player evaluations from a leading NFL analytics company who use a combination of statistics and expert evaluation. In addition, it will investigate what features are significant in the evaluation of a position. Data for the dissertation is obtained from multiple online sources - Pro Football Reference and Pro Football Focus (the the NFL analytics company). These data sets are combined and analysed before applying six different approaches to the problem. The use of Neural Networks (both Single and Multi Layer) as an approach is evaluated against the other approaches of Support Vector Regression (SVR), Linear Regression, Decision Trees and XGBoost. They will be evaluated using accuracy, root mean squared error and the p-value from a t-test. Wrapper methods of Sequential Feature Selection and Permutation Importance are both used to discover relevant features. SVR was the best performing approach with 74% accuracy for QB, 76% accuracy for WR and 59% for RB. Both XGBoost and the Neural Network implementations performed well in comparison. The relevant features that were uncovered fell into two distinct categories. First is a measure of the ability of the player to make an impact on the game when they are involved and receive the ball. The second is a highlight of the importance of solid foundations and basics