AYRIK GİZLİ MARKOV MODELİNİ KULLANARAK SEGMENTASYON YOLUYLA TREND TESPİTİ

There are two models that benefit from the concept of hidden variables and are used frequently in practice. These are hidden Markov model and Markov switching model. Although there are quite a number of regime switching studies which applied the Markov switching model in the field of econometrics, studies that use the hidden Markov model for examining the regimes in econometric time series are sparse. In this study, we apply discrete hidden Markov model to four high-frequency time series and show that although transformed discrete data have to be used in the model structure, the model identifies two or more regimes quite well. It is concluded that the discrete hidden Markov model is defining regimes effectively, thereby, can also identify and segment trends.