SYNTHESIS TRENDS OF FORECASTING USING INDUCTIVE MODELING METHODS

Abstract

Modern development of computer technology and the possibility of implementing calculations in parallel allow to solve increasingly large-scale problems of numerical modeling. The development of multiprocessor computing and parallel computing makes it important to solve problems of optimization analysis. The optimization analysis is based on the mass solution of inverse problems when the defining parameters of the considered class of problems change in certain ranges. Thus, calculations of not only direct problems where it is necessary to model the phenomenon at the known initial data, but also calculations of inverse problems where it is necessary to define on what defining parameters there is this or that phenomenon become more and more demanded. This formulation requires multiple solutions of direct problems and solving the problem of optimization analysis and construction of predictive trends. Sets of multidimensional parametric data in the paper are considered as numerical solutions of the optimization problem. The construction of predictive trends is implemented on the basis of the group method of data handling as a direction of induction modeling. The methodology of visualization of results of calculation of parametric functions is realized. The scheme of Data Mining with application of methods of visualization by means of the Matlab software environment is described