McAnulty College and Graduate School of Liberal Arts
John C. Kern
Posterior distribution, prior distribution, mcmc sampling, gibbs sampling, credibility interval
This thesis explores variations on a Bayesian regression model used to estimate the mean box length of a random knot as a function of the number of edges of that knot. Specifically, this research recognizes uncertainty in box length variance and compares the resulting inference with that based on an approach that does not recognize such uncertainty. The Bayesian model is then shown to allow straightforward inference on the crossing location of two population regression lines.
Bilir, S. (2008). A Comparison of Bayesian Regression Models Applied in Knot Theory (Master's thesis, Duquesne University). Retrieved from https://dsc.duq.edu/etd/318