Toolkit for variative forecasting of time series

Abstract

In this paper, we describe structure, features, and application examples of the toolkit for variative forecasting of time series (VarForecasts). A set of original algorithms for identification and forecasting of univariate discrete time series is implemented in the system. The toolkit allows for a comprehensive study of time series on the basis of ideas and methods of variative modeling. VarForecasts can be used in different subject areas for analysis and forecasting of discrete time series. Algorithms presented in toolkit have been tested on data on infectious diseases in cities of Russia and hydrological data on inflow of the Ob river. Obtained results allow conclusion on the prospects of the VarForecasts as a means of analysis and forecasting of discrete time series.