RESEARCH ON THE SENSITIVITY OF MULTI- CRITERION EVALUATION METHODS/SENSIBILITÄTSUNTERSUCHUNGEN ZU METHODEN DER MEHRKRITERIELLEN ENTSCHEIDUNGEN
Investigations into the multi-criterion evaluations have been performed for different purposes for more than 30 years. Nevertheless, up to now there are no common rules how to apply multi-criterion methods of evaluation and how to interpret their results. But the solutions of the problem must be found. The investigations are based on the joint programme LEVI developed by the VGTU and the Leipzig HSTEC. In accordance with the programme, for solving the problem a matrix is created in which all solution variants are evaluated by the same criteria. For the transformation of initial data different methods are used. Besides, the problem solution may be oriented to a choice of a variant and the determination of an optimal sequence or to the determination of the rational behaviour balance of two adversely interested groups. The maximisation, minimisation and mixed problems are distinguished. In case of the last mentioned problems the meanings may be unsuitable and expressions about the maximisation or minimisation too strong. When solving the problem of maximisation by different transformation methods, similar curves are obtained and the results are not dispersed heavily. In case of the minimisation, when solving according to different transformations, a considerable deviation of results has been noticed. The investigations also disclosed that solutions may vary depending on the optimisation goal. Dispersion of results in case of minimisation is stronger than in case of maximisation. When analysing different transformations, the vectorial transformation is regarded as “neutral” and there is no sense to include it into comparison. In case of the minimisation problem, linear transformations should be investigated in a more detailed way.