Bayesian model determination for binary time-series data with applications (Q5941335): Difference between revisions

This article considers the Bayesian analysis of binary time-series data. Continuous latent random variables are introduced to develop a regression model containing some exogenous variables and past experiences expressed by an autoregressive model. The order of the autoregressive component is treated as a parameter, instead of being fixed in advance. By using some reversible Markov chain Monte Carlo technique proposed by \textit{P. J. Green} [Biometrika 82, 711-732 (1995; Zbl 0861.62023)], the order of the autoregressive component is determined, and the marginal posterior density of each parameter and the multi-period predictive probabilities are, respectively, estimated. Simulation is provided to examine influences on the order determination, via the size of the binary time series. Also, one set of real data, the Six Cities study of the health effects of air pollution, is applied to illustrate the usage of the proposed model. Some interesting results about maternal smoking and child's wheezing are interpreted.