Abstract: Earlier, in the framework of our previous papers, we introduced a criterion for comparing interval alternatives, which was called the assurance factor. It is based on the balance of chances that hypothesis on preference for one of the compared alternatives is true and the risks associated with the possible truth of the opposite hypothesis. The criterion is dimensionless that is not similar to other criteria used for this purpose. In this paper this criterion and its calculation method are proposed to use for analysis of the most important from a practical point of view problems of forecasting the expected effectiveness of the individual alternatives. Numerical procedures have been developed to calculate the values of the assurance factor in the case of arbitrary distribution on the analyzed interval alternative. Analytical expressions were also found for the uniform and triangular distributions. Comparison of the results was produced for evaluating the expected effectiveness of interval alternatives on the proposed criterion and on base of other evaluation criteria (mean value, Hurwicz criterion). The arguments presented in favor of adequacy of the proposed criterion for problems of forecasting both expected efficiency of interval alternatives and comparison of alternatives on preference. The possibility harmonizing of results comparing alternatives and evaluating their expected efficiency based on Hurwicz criterion and on the assurance factor was considered. Relation of the assurance factor with "pessimism - optimism" coefficient was received. On this basis we analyze the question concerning substantiation of the choice of Hurwicz criterion values. Using ordinal scales in the process of comparing alternatives and evaluating their expected effectiveness on base of the assurance factor is proposed.

Keywords: estimating expected efficiency of interval alternatives, criterion estimating efficiency, comparing different criteria of efficiency

ACM Classification Keywords: H.1.2 Human information processing. G3 Distribution functions. I.2.3 Uncertainty, “fuzzy”, and probabilistic reasoning

Link:

Evaluating Expected Effectiveness of Interval Alternatives

Gennady Shepelev, Mikhail Sternin

http://www.foibg.com/ijita/vol21/ijita21-03-p06.pdf