DECISION SUPPORT SYSTEM FOR EVALUATING RANKINGS OF ALTERNATIVES IN MULTICRITERIA CHOICE

The paper describes the software complex for the construction of a set of rankings and their scalar and vector estimates in the case of multicriteria choice. The term «ranking» is similar to term «path» on directed graph. Each criterion in a real problem can be a binary relation on a set of alternatives («alternative» is the same as «vertices» on graph), which are proposed for choice, and the binary relation can correspond to a weighted digraph. The weights could be interpreted as estimates of one criterion, describing preferences and valuations of every pair interrelation in the set of pairs (as binary relation provides pairwise comparisons). Every ranking can be given as set of pairs – as strict linear fuzzy order described by some matrix of pairwise comparisons. The ranking can be matched with several matrices of criteria and, leaning on several matrices, the several estimates can be calculated. Besides, every given ranking could be estimated by the aggregated relation, which is minimally remoted from all the criteria relation, constructed with the criteria matrices. Using the software system, you can build all possible rankings and calculate the values of the characteristics of each ranking, with respect to the matrix of each criterion. Based on the obtained vector estimates of paths (or «rankings»), optimal sets of paths can be calculated. Elements of such sets are considered to be the best in terms of taking into account the estimates of each criterion. The functioning of the created decision support system is demonstrated using the example of an applied problem with a road network, where weights of the given weighted oriented graphs are considered to be the estimates of suitedness of every road (every pair of vertices) by some criterion. Based on the obtained estimates (vector and scalar characteristics), the best paths through the vertices are selected.