time series model
Intellectual analysis of the behavior of complex systems
For the analysis of the behavior of complex systems in the last time are actively developing models and methods directions of intelligent analysis of time series Time Series Data Mining) in particular on the basis of neural network and fuzzy models. However, the neural network models do not have formal representation, and also do not provide for the linguistic interpretation of results the analysis of time series, which limits their application.
The idea of fuzzy modeling the behavior of complex systems for time series is based on the basic model of the fuzzy process, which is called a fuzzy time series (FTS). Methods of analysis FTS showed their productivity in predicting the behavior of complex systems in transport, ICT, in the field of electric power industry, education, economy, ecology and engineering. The decision of problems of the analysis, modeling and forecasting of the fuzzy trends, detailed in the framework of the fuzzy process with fuzzy increments, for a number of subject areas will ensure the possibility of the extraction of useful qualitative information in terms of fuzzy trends, complementing the stages of the quantitative analysis and the analysis of time series.
In this article addresses the question about the correlation of quantitative (statistical, neural network models) and qualitative (fuzzy time series models and fuzzy trends models). From the methodological point of view of the introduction of the new object, fuzzy trends, raises the question of its place in the system of well-known objects of the analysis of time series. It is proposed granular presentation of the system of objects of the analysis of time series as the most simple decision of the indicated questions at the content level.