A hidden Markov regime is a Markov process that governs the time or spacedependent distributions of an observed stochastic process. In the studied model, a hidden Markov chain governs the distribution of a mixed autoregressive process. This paper deals with simultaneous recursive parameter estimation and reconstruction of the hidden Markov chain. The authors use a MAP estimate for the reconstruction of the Markov chain together with a recursive EM-algorithm for the parameters. Simulations are made in models with two and three Markov states. An attempt has been made to fit real data from a clock striking twelve to the model. This paper also includes an example of robust estimation for data with clustered outliers, where one Markov state represents the outliers among the observations.
展开▼