| |
| author |
Lloyd Goldstein
| | title |
What Makes a Good Quarterback? Analysis of Longitudinal NFL Data Using Latent Variable Clustering Methods
| | abstract |
This thesis uses latent variable clustering methods to analyze longitudinal NFL
quarterback data in a previously unexplored way. The main method used in this work is Latent
Class Analysis (LCA) and its longitudinal extension Latent Transition Analysis (LTA). These
methods use dichotomous, longitudinal performance data to create clusters of quarterbacks.
Football performance data for 22 quarterbacks from 2012 to 2015 is used for analysis. The
results of this clustering are then compared with results generated by other clustering methods.
They are also compared with conventional football analysis from reputable websites such as
ESPN. The results of the latent variable clustering methods are generally in line with those
generated by other clustering methods. They also reflect conventional football wisdom quite
accurately, but with a bit more specificity. For this football data set, latent variable clustering
methods are effective and interpretable methods of quarterback classification.
| | school |
The College of Liberal Arts, Drew University
| | degree |
B.S. (2020)
|
| advisor |
John Kettenring
|
| full text | LGoldstein.pdf |
| |
