The methodical approach to forecasting an uneven time series

Russkov O. V., Voronkov I. A., Saradgishvili S. E.

The sources of time series unevenness are shown. The existing methods of time series forecasting are considered. Their advantages and disadvantages are determined. A methodical approach to forecasting an uneven time series is described. The principal applicability of the described forecasting method is proved by the example of forecasting a power consumption of an industrial enterprise. The software model based on the described method has been tested. The superiority of described model results over results of autoregressive and neural network models is shown.

Keywords:time series, forecasting, unevenness, game theory, system approach