A0976
Title: Bayesian dynamic clustering factor models with regressors
Authors: Tsering Dolkar - Virginia Tech (United States) [presenting]
Marco Ferreira - Virginia Tech (United States)
Allison Tegge - Virginia Tech (United States)
Hwasoo Shin - Virginia Tech (United States)
Abstract: A novel class of Bayesian dynamic clustering factor models with regressors are proposed. This new class of factor models is useful for the analysis of multivariate longitudinal data on a sample of subjects. It is assumed that at each time point, each subject belongs to one of many clusters, and the subject may move to another cluster at the next time point. Further, the probability of a subject moving from one cluster to the other clusters depends on regressors. These regressors may include changes in individual-level psychosocial factors from one time point to the next time point. A Markov chain Monte Carlo algorithm is developed to explore the posterior distribution of the unknown quantities. The usefulness of the novel framework is illustrated with an application to health data.
