Information Technology of Data Mining and Forecasting of Nonlinear Non-Stationary Processes

Abstract

The article is devoted to the actual task - the information technology development for the data intellectual analysis and nonlinear non-stationary processes (NNP) forecasting. The study object are non-linear non-stationary processes in ecology, economics and finance. The necessity of development of information technology for data mining and forecasting of nonlinear non-stationary physical processes of different nature is grounded in order to increase the adequacy of mathematical models of nonlinear non-stationary processes and improve the quality of prediction estimates that are calculated using available models. The rules of its application are described. The mathematical apparatus used in the composition of information technology for solving problems of various nature data analysis is presented for the purpose nonlinear nonstationary processes forecasting. The results of information technology application for forecasting of such processes are resulted. The linguistic modeling method is described in detail. The content and mathematical formulation of the task of finding linguistic images of time series is formulated. The concept of linguistic modeling was introduced and the method of applying this approach was gradually described to solve the problem of prediction of nonlinear non-stationary physical processes of different nature. Also, for a more complete consideration of the roblem, additional subtasks were introduced, their mathematical statements were described, and examples of their execution were given. In the scientific researches course, nonlinearities and nonstationary aspects types were identified in modern environmental, financial and economic processes. NNP classification is made, the existing information technologies for the research of NNP analysis is carried out; processes for research are selected; the necessary statistics are collected; the forecasting function is constructed and estimates of forecasts for the selected processes are calculated. The the practical verification results of the use of the developed information technology for prediction of nonlinear non-stationary processes with the help of the linguistic modeling method are presented. The results of the comparative analysis of the developed method adequacy with some existing ones are analyzed and the quality of the obtained predictive values is analyzed. The analysis showed the high adequacy of the results obtained with the help of the developed information technology and high quality of the received forecasts.