A1725
Title: Bayesian graph estimation under causal vector autoregressive time series
Authors: Anindya Roy - U.S. Census Bureau (United States) [presenting]
Abstract: Multivariate time series data are routinely collected in many application areas. Although stationarity, causality, and invertibility are very useful modelling assumptions for time series data, methodological developments are limited under these assumptions for multivariate time series. These properties are achieved for a high dimensional Gaussian vector autoregressive (VAR) time series while modelling the graphical dependence structure among the variables. A new parameterization is proposed that models the marginal precision matrix as well as the VAR dynamics and develops related computational methodologies. Theoretical consistency properties of the method are studied and its performance is illustrated through simulation and real data applications.
