approach is discussed. It’s showed that to compare two poly-interval estimations (PIEs) we may compare their different mono-interval realizations, i.e. different pairs of mono-interval estimations from compared PIEs, using a Monte-Carlo? family method. To compare mono-intervals people often use equivalent point estimations of intervals received on base of Hurwicz’s criterion or similar criteria. We offer a method of direct comparison of mono-interval estimations that interprets the problem as a problem of decision-making. We suggest distinguishing a favorable zone (zone where one interval in comparable pair may be more preferable than another), a zone of uncertainty (zone where compared intervals are equivalent to each other) and an unfavorable zone. Preferences of the decision maker during analysis of relative widths of the zones allow analyze the problem. The results of direct comparison are compared with results received on the base of equivalent point estimations. The arguments are given that method of direct comparison of intervals is more adequate to decision-making tasks. Relationship between numbers of preferable, equivalent and no preferable comparisons of mono-intervals from PIEs gives an information that support process of decision making during comparison of poly-interval estimations by decision maker.

Keywords: generalized interval estimations, comparison of intervals, decision support

ACM Classification Keywords: H.4.2 Types of Systems – Decision support; G.1.0 Numerical Analysis - Interval Arithmetic, G.3 Probability and Statistics – Distribution functions

Link:

СРАВНЕНИЕ ПОЛИИНТЕРВАЛЬНЫХ ОЦЕНОК В МЕТОДЕ ОИО

Михаил Стернин, Геннадий Шепелёв

http://foibg.com/ibs_isc/ibs-10/ibs-10-p11.pdf