B1656
Title: Using covariate informed partition models to identify subpopulations via curve clustering
Authors: Garritt Page - Brigham Young University (United States) [presenting]
Abstract: Studies are considered that measure functional output on a collection of experimental units or subjects. The main objective is to flexibly model individual curves while simultaneously assigning subjects to clusters based on curve shape such that the resulting partition is scientifically meaningful. Additionally, we seamlessly incorporate covariate information to the formation of curve clusters by employing a covariate informed partition model. Once clusters are connected with covariate information, we show how this information can be used to carry out less expensive/invasive diagnosis or prediction of a subjects future behavior.
Journal