CONSTRUCTION OF MULTI-CRITERIA SOLUTIONS TO SELECTION PROBLEMS

Abstract

An approach to solving the problems of multi-criteria selection with the determination of suitable applicants based on the values of a number of criteria is considered. Known approaches to the solution of these problems do not always allow determining reasonable solutions with full consideration of the features of the selected criteria and their evaluation conditions. Comparison of evaluations by criteria often takes place from subjective positions without proper justification, which affects both the results of the selection and the possibilities of their practical application. Known multi-criteria selection procedures use fixed measurement scales to compare values, which in practical situations may not correspond to real ratios of indicators. Another significant shortcoming of these procedures is ignoring the presence of non-linear dependencies between evaluations by criteria and attributes of decisions made in different ranges of their values. The above indicates the possibility of obtaining inadequate evaluations of decisions with corresponding negative consequences. The paper proposes an approach to the construction of solutions to multi-criteria selection problems based on the definition of applicants by ordering them according to the degree of approximation to the option with the best ratios of evaluations according to the selected performance criteria based on iterative concessions according to the specified criteria. Applicants are compared on the basis of the ratios of assessments according to the agreed criteria, the definition of which is provided by a set of mathematical models that reproduce the dependence of the values of individual criteria on the attributes of the applicant decisions. The formation of a sequence of candidates for solutions is based on the results of comparisons according to individual criteria, which allows to determine alternative solutions and evaluate their effectiveness with the construction of a Pareto subset. An example is considered that reproduces the processes of choosing alternative options for building a virtual data processing system based on the characteristics of virtual machines from providers of cloud processing technologies. The obtained research results indicate the possibility of using the proposed approach as part of decision-making support subsystems to solve the problems of operational management of dynamic service and production processes.