Predictive modeling of professional figure skating tournament data

This thesis examines various stochastic models for modeling the performance scores earned by competitors in figure skating competitions. Using the results from 107 professional figure skating tournaments over a 5-year period, a reasonably par- simonious model relating performance on the initial (short program) segment to performance on the final (long program) segment is derived. Tests of fit and checks for model validity and accuracy are also conducted to show that the model is ade- quate.