Automation of the process of building a time series model based on formalized expert knowledge

Abstract

The article is devoted to the problem of automation of the process of constructing a time series model of the autoregressive type of integrated moving average - Box-Jenkins model. The process of automation is carried out on the basis of the formalized expert knowledge about the joint behavior of graphs of the dependence of the graphs of the autocorrelation and partial autocorrelation functions of a time series values on the magnitude of the delay between sections of the process under study. We used the information about heuristically established relationships between the values of the structural parameters values of the Box-Jenkins model and the peculiarities of behavior changes of the autocorrelation and partial autocorrelation functions. A feature of the proposed method of solving the problem is the use of algorithmically formalized procedures for processing expert knowledge by means of mathematical apparatus of fuzzy set theory.