Short-term Forecasting Information System

Abstract

The paper considers the information system through which the process of automating the work of the prediction filter, which is constructed as a tool for the computational and predictive process, has been considered. Synthesized models of varying complexity due selected Kolmogorov-Gabor polynomial 2 degrees. The reasons for choosing a polynomial, creating a system of conditional equations and the number of equations in a normalized system based on the principle of their formation are given. The results of calculations trained models allowed not only to determine the forecast error and to analyze the behavior of the entire process as a whole. By calculating the mean square error of the forecast, we can determine the most optimal model required for checking the predicted sequence. They explained that the lower the percentage error, the better the trained model from all the predicted sequences. All information on the calculations, the selection of the input array, the obtained values of models and the type of the synthesized model, as well as the calculation of the forecast error, is stored in the historical data of the automated information system. The information system has a comprehensive program algorithm for predictive filter in the form of a user-friendly interface.