UDC 004.855.5
MULTI-OBJECTIVE OPTIMIZATION FOR THE FORECASTING MODELS ON THE BASE OF STRICTLY BINARY TREES
L. A. Demidova, PhD (technical sciences), full professor, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it.
N. N. Astakhova, post-graduate student, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it.
The problem of multi-objective optimization dealing with the development of forecasting models on the base of strictly binary trees is considered. The aim is to improve the search characteristics of a modified clonal selection algorithm which is applied for the development of forecasting models on the base of strictly binary trees by means of involvement in the process of selection of best forecasting models of two model quality indicators: the affinity indicator and the indicator of the tendencies' discrepancy. The accounting of two indicators of a forecasting model quality is carried out with the use of the notion «Pareto-set» which is applied in the formation process of new populations of forecasting models in a modified clonal selection algorithm. During the formation of new population of forecasting models for maintenance of its high variety it is offered to consider values of the crowding-distance of forecasting models. The results of experimental studies confirming the efficiency of the offered approach to improvement of the search characteristics of a modified clonal selection algorithm are given.
Key words: time series, forecasting model, strictly binary tree, modified clonal selection algorithm, genetic algorithm, multi-objective optimization, Pareto-dominance, crowding-distance.