A1585
Title: A comparison of parsimonious families of hidden Markov models for multivariate longitudinal data
Authors: Mackenzie Neal - McMaster University (Canada) [presenting]
Paul McNicholas - McMaster University (Canada)
Abstract: The popularization of hidden Markov models (HMMs) for analyzing multivariate longitudinal datasets wherein estimating state switching of subjects is desirable has resulted in over-parameterization issues. Thus, parsimonious HMMs are essential for the analysis of longitudinal datasets. Commonly, parsimony is introduced by imposing a series of constraints on decomposed covariance matrices. Two families of HMMs arising from the modified Cholesky decomposition and the eigenvalue decomposition on various longitudinal datasets are introduced and compared.
