Mixture Multi-Step-Ahead Prediction

0450479 - ÚTIA 2016 RIV GR eng C - Konferenční příspěvek (zahraniční konf.)
Nagy, Ivan - Suzdaleva, Evgenia - Mlynářová, Tereza
Mixture Multi-Step-Ahead Prediction.
Proceedings of the 16th conference of the Applied Stochastic Models and Data Analysis (ASMDA) International Society. Piraeus: ISAST: International Society for the Advancement of Science and Technology, 2015, s. 727-738. ISBN 978-618-5180-05-8.
[The 16th conference of the Applied Stochastic Models and Data Analysis (ASMDA) International Society. Piraeus (GR), 30.06.2015-4.07.2015]
Grant CEP: GA ČR GA15-03564S; GA MŠMT 7H14004; GA MŠMT(BE) 7H14005
Institucionální podpora: RVO:67985556
Klíčová slova: mixture model * mixture prediction * recursive mixture estimation * dynamic pointer * active component
Kód oboru RIV: BB - Aplikovaná statistika, operační výzkum
http://library.utia.cas.cz/separaty/2015/ZS/suzdaleva-0450479.pdf

The presented paper deals with a task of the multi-step prediction with mixture models under Bayesian methodology. The main contribution of the paper is a recursive prediction algorithm for mixtures with the dynamic switching model. The proposed algorithm is based on construction of the weighting vector predicting the active component and its combination with data predictions from components. With the help of illustrative examples the paper compares the results with those obtained for the mixture prediction with the static switching model.
Trvalý link: http://hdl.handle.net/11104/0251934