Parsimonious models of multivariate binary time series: statistical estimation and forecasting

Abstract

This paper is devoted to parsimonious models of multivariate binary time series. Consistent asymptotically normal statistical estimators for the parameters of proposed parsimonious models are constructed. Algorithms for statistical estimation of model parameters and forecasting of future states of time series are presented. Results of computer experiments on simulated and real statistical discrete-valued data are given