Giacomo’s dissertation, supervised by Gareth Roberts of the University of Warwick, showcases the development of innovative statistical methods inspired by a difficult applied problem. The motivating application is complementary clustering of ancient Anglo-Saxon settlements, in which each cluster is comprised of settlements with different types. The thesis develops a Bayesian Random Partition Model, and relates intractability of the posterior distribution to results from the Computational Complexity Theory. It then proposes a novel and efficient set of MCMC methods based on insights into the underlying theory. These methods produce very satisfying results for the particular application, and are applicable more general.
Each year the Savage Award, named in honor of Leonard J. “Jimmie” Savage, is given to two outstanding doctoral dissertations in Bayesian econometrics and statistics, one in Theory and Methods and another in Applied Methodology.