Premium
Bayesian multiple change points and segmentation: Application to homogenization of climatic series
Author(s) -
Hannart A.,
Naveau P.
Publication year - 2009
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2008wr007689
Subject(s) - jump , bayesian probability , series (stratigraphy) , markov chain monte carlo , computer science , algorithm , segmentation , markov chain , jump process , reversible jump markov chain monte carlo , mathematics , mathematical optimization , artificial intelligence , machine learning , biology , paleontology , physics , quantum mechanics
In this paper, we describe a new multiple change point detection technique based on segmenting the time series under study into subsequences. These segments correspond to the episodes that are likely to contain a unique jump. They are found by applying Bayesian decision theory through the minimization of simple cost functions. All calculations can be performed explicitly, without falling back on Markov chain Monte Carlo methods and resulting in particularly light implementation. Through prior distributions derived from a stochastic renewal process description of jump occurrences, expert knowledge of jump amplitude and return period is also introduced in our decision process. Comparison to several multiple change point methods on simulated series lead to similar or better performance, achieved at lower computational cost.