Dynamic Models for Multivariate Time Series of Counts, with Applications 3:30 PM – 4:30 PM, Thursday November 7th, 2013 Marrs McLean Science Building, Room GL 51 Nalini Ravishanker, Ph.D. Department of Statistics, University of Connecticut Abstract Time series counts data occurs in many disciplines and there is considerable interest in developing accurate methods for modeling and prediction. This talk describes dynamic models for univariate and multivariate time series of counts, incorporating time dependence and dependence between the components of the series. Dynamic generalized linear models and nonlinear models, including a hierarchical setup to handle subject-specific estimation, are described and a Bayesian framework for inference is discussed. The approach is illustrated using an application from ecology for modeling gastropod abundance. A reception with refreshments will be held from 3:00 PM – 3:30 PM in the lobby of the Statistical Science Department, on the first floor of Marrs McLean Science.