Декомпозиция и прогнозирование временных рядов с долговременными корреляциями
Тебуева Фариза Биляловна, Перепелица Виталий Афанасьевич, M. Yu. Kabinyakov · Известия Южного федерального университета. Технические науки · 2013
N article the technique of decomposition of time series with long-term correlations is offered. Feature of time series of this class is non-obedience to the normal law, lack of a visible trend and non-performance of a condition of independence of values of a row. For receiving trend components Hurst's modified algorithm, allowing to allocate trend pieces in a temporary row is used. Forecasting is carried out on the basis of cellular and automatic model in which the configuration for the forecast decides on memory by depth of memory of series.
Cited by 0 works.