Detecting and Estimating the Location of Shifts in the Transition Probabilities of Two-State Markov Chains.
CASE WESTERN RESERVE UNIV CLEVELAND OHIO DEPT OF MATHEMATICS AND STATISTICS
Pagination or Media Count:
K independent and identically distributed two-state Markov chains are subjected to shifts in their transition probabilities, which occur at unknown time points. Two detection and estimation procedures, concerning the location of the shift, are developed in a Bayesian framework for an at-most-one-change model. Estimates of the operating characteristics of these procedures have been obtained by simulating such Markov chains. The simulation results indicate that the proposed procedure, which is based on comparison of posterior risks, could be very effective. Author
- Operations Research