Speaker: Garritt Page, Professor, Statistics, Brigham Young University
Title: Informed Random Partition Models
Abstract: Model-based clustering is a powerful tool that is often used to discover hidden structure in data by grouping observational units that exhibit similar response values. Recently, clustering methods have been developed that permit incorporating an ``initial'' partition informed by expert opinion. Then, using some similarity criteria, partitions different from the initial one are down weighted, i.e. they are assigned reduced probabilities. These methods represent an exciting new direction of method development in clustering techniques. We add to this literature a method that very flexibly permits assigning varying levels of uncertainty to any subset of the partition. This is particularly useful in practice as there is rarely clear prior information with regard to the entire partition. Our approach is not based on partition penalties but considers individual allocation probabilities for each unit (e.g., locally weighted prior information). We illustrate the gains in prior specification flexibility via simulation studies and an application to a dataset concerning spatio-temporal evolution of PM10 measurements in Germany.
Short Bio: Garritt L. Page is Professor in the Department of Statistics at Brigham Young University. He started his academic career at Duke as a postdoc in 2009 and then at Pontificia Universidad Catolica de Chile as an Assistant Professor in 2011. He did doctoral work at Iowa State University under the direction of Stephen Vardeman. Garritt L. Page's research interests are Bayesian (non)parametrics with focus on model-based clustering and random partition models in addition to functional data, spatial confounding, and sports analytics. His work has appeared in journals including the Journal of the American Statistical Association, Annals of Applied Statistics, Journal of Computational and Graphical Statistics, and Bayesian Analysis.
