ISSN 1991-2927
 

ACP № 4 (62) 2020

Author: "Mkrtych Sarkisovich Tonerian"

Tatiana Vasilievna Afanaseva, Ulyanovsk State Technical University, Doctor of Engineering; Associate Professor, Deputy Head of Information System Department at Ulyanovsk State Technical University; graduated from the Faculty of Radioengineering at Ulyanovsk State Technical University; an author of articles and monographs in the field of the intellectual analysis of time series. [e-mail: tv.afanasjeva@gmail.com]T. Afanaseva,

Aleksei Andreevich Sapunkov, Ulyanovsk State Technical University, Post-graduate Student of the Information System Department at Ulyanovsk State Technical University; graduated from the Faculty of Information Systems and Technologies at Ulyanovsk State Technical University; an author of articles in the field of the intellectual analysis of time series. [e-mail: sapalks@gmail.com]A. Sapunkov,

Mkrtych Sarkisovich Tonerian, Ulyanovsk State Technical University, Post-graduate Student of the Information System Department at Ulyanovsk State Technical University; graduated from the Faculty of Information Systems and Technologies at Ulyanovsk State Technical University; an author of articles in the field of the intellectual analysis of time series. [e-mail: mkr73@yandex.ru]M. Tonerian

The Two-stage Algorithm of Choosing the Fuzzy Model for Time Series Firecasting 000_8.pdf

The article deals with a two-stage algorithm for the best time-series forecasting model based on the assessment of the model adequacy with the use of behavior and accuracy. For testing, the authors use time series that have been exploited at the International Time Series Forecasting Competition IFAS-EUSLAT in 2015 ([http://irafm.osu.cz/cif/main.php]). Database of this Competition includes 91 numerical time series of different length, tendency, and data reading frequency. Time series values depicted the dynamic of parameters reported from banking area, social networks, and medicine. Three models based on the fuzzy time series concept have been used for time-series forecasting. In order to choose the best model, the two-stage algorithm based on the comparison of time series and model tendencies has been proposed. In addition to the already known quality criterion, the new ones are also exploited in the algorithm. In the conclusion, the results obtained are discussed and the effectiveness of the suggested algorithm is demonstrated.

Fuzzy tendency, fuzzy time series, forecasting, linguistic description.

2015_ 4

Sections: Information systems

Subjects: Information systems.


© FRPC JSC 'RPA 'Mars', 2009-2018 The web-site runs on Joomla!