Dipartimento di Scienze Statistiche

Giovedì 29 febbraio, ore 15.00   Aula 24 - quarto piano (Ed. CU002)   Latent Factor Multivariate Integer-Valued AR Processes   Refik Soyer (Department of Decision Sciences, George Washington University, US)   Abstract: We introduce a new class of multivariate integer-valued autoregressive (INAR) models based on the notion of a common random environment. Dependence among the components of the multivariate time series is induced via the common environment that follows a Markovian evolution. The proposed framework provides us with a dynamic multivariate generalization of the univariate INAR processes. We develop a Markov chain Monte Carlo method as well as a particle learning algorithm for Bayesian inference.  We consider an extension of the model to handle zero inflated time-series data and illustrate the proposed class of models using actual multivariate count data and discuss their predictive performance. (Joint work with Di Zhang, Expedia, US and Hedibert Lopes, INSPER, Brazil)