Open AccessHierarchical Bayesian segmentation for piecewise stationary autoregressive model based on reversible jump MCMC
Author(s) -
Suparman Suparman
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1321/2/022067
Subject(s) - autoregressive model , markov chain monte carlo , reversible jump markov chain monte carlo , bayesian probability , estimator , star model , algorithm , piecewise , bayesian inference , mathematics , jump , series (stratigraphy) , computer science , metropolis–hastings algorithm , time series , statistics , autoregressive integrated moving average , mathematical analysis , paleontology , physics , quantum mechanics , biology
This paper aims to decompose time series data in segments where many segments are unknown. The data in each segment is modeled as a stationary autoregressive where the model order is unknown. The model parameters include the number of segments, the location of segment changes, the order of each segment, and the autoregressive coefficients of each segment. The Bayesian method is used to estimate parameters, but Bayesian estimator cannot be calculated analytically. The Bayesian estimator is calculated using the reversible jump Markov chain Monte Carlo algorithm. The performance of the algorithm is tested using synthesis data. The simulation results show that the algorithm estimates the model parameters well.