MATHEMATICAL MODELING

Adaptive structural forecasting methods

J. A. Dorofeyuk

V. A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences, Moscow

Abstract: The analysis and forecasting problem in multivariate control system is described. The object structure of investigated system is assumed to vary over time. The adaptive algorithm, which is able to trace such structure variations, is developed. The integral object description and adaptive forecasting model allow to improve substantially the efficiency of management decision results.

Keywords: classification data analysis methods, adaptive automatic classification algorithm, adaptive forecasting model, class number selection optimization.