Identification and Accounting of Aperiodic Anomalies in Time Series with Cyclic Components
Valeriya Morozova, Valery Menshikh · 2023
This article is devoted to solving the problem of studying time series with aperiodic point anomalous changes in the level values of the series. In addition, these series contain a cyclic component. The use of classical mathematical methods leads to obtaining models with unacceptable accuracy. The article developed methods for modeling time series with anomalous changes. The developed methods are based on the detection and exclusion of anomalous changes in the values of the series levels. In this case, the anomalous values of the levels of the series are subdivided into structural and parametric ones. An indicator method for detecting structural anomalies has been developed. And the detection of parametric anomalies is based on the assumption of the normality of the law of distribution of the values of the cyclic component with point exceptions of the levels of the series. To solve the problem, cyclic and trend components were identified at intervals of non-anomalous changes in level values.