A system approach to decision support on basis of hierarchical and network models

Abstract

A methodology and a system approach to decision support based on hierarchical and network models are proposed, which include methods for estimating and improving the consistency of pairwise comparison matrices (PCM), calculating fuzzy local weights for model elements based on fuzzy PCM, calculating confidence intervals for local weights, finding aggregated weights and aggregated belief functions for model elements, estimating the sensitivity of the results, tools for analyzing the effectiveness of the methods and an expert evaluation modeling system. A decision support system is built on basis of the proposed methodology. The proposed methodology makes it possible to increase the reliability of solutions of practical weakly structured decision-making problems by using more efficient methods at each of the stages of solving the problem. Reliability of results of application of the methodology while solving a practical decision-making problem can be assessed by means of sensitivity analysis using the stability indices of the computed local and global rankings of the decision alternatives.