Biclustering longitudinal trajectories through a model-based approach

This work introduces a model-based biclustering approach for discrete multivariate longitudinal data. The proposed model considers a finite mixture of generalized linear models to cluster units and, within each mixture component, a flexible and parsimonious parameterization of the corresponding canonical parameter to cluster variables evolving in a similar manner across time. Model parameter estimates are obtained through an Expectation Maximization (EM) type algorithm and the performance of the proposed model are shown on both simulated and real dataset.