Centralni deo Analitičkog hijerarhijskog procesa (AHP), metoda za podršku procesa donošenja odluka, jeste poređenje u parovima elemenata hijerarhije i formiranje odgovarajućih lokalnih recipročnih numeričkih matrica iz kojih se nekim od matematičkih postupaka određuju težine poređenih elemenata.
Matrice i izračunate težine elemenata nose informaciju o konzistentnosti donosioca odluka koja se može meriti na razne načine, direktno i indirektno. U standardnom AHP koristi se stepen konzistentnosti (CR), a od drugih mera najčešće je u upotrebi totalno Euklidsko rastojanje (ED).
Ako ima K donosilaca odluka koji tretiraju isti problem, tada na svakom nivou hijerarhije umesto jedne lokalne matrice ima K takvih matrica i isto toliko skupova težina elemenata. Sinteza individualno dobijenih težina u grupne težine vrši se na razne načine, ali je problem kako se odnositi prema različitoj konzistentnosti više donosioca odluka.
U radu se taj problem tretira preko izabrana tri parametra performanse donosioca odluka, definiše se postupak analize i na jednom primeru ilustruje kako se mogu identifikovati donosioci odluka koje treba uključiti pri sintezi individualnih u grupne težine.
Na individualnom nivou korišćene su mere konzistentnosti CR i ED, a grupna konzistentnost tretirana je pomoću Spirmanovog koeficijenta korelacije rangova.
Ključne reči: AHP, donosilac odluka, grupa, konzistentnost
Bojan SRĐEVIĆ, Kosana SUVOČAREV, Zorica SRĐEVIĆ
Univerzitet u Novom Sadu, Poljoprivredni fakultet, Departman za uređenje voda
E-mail: (bojans, suvocarev, srdjevicz)@polj.ns.ac.yu
Ceo sadržaj ovog naučnog rada možete pročitati u Vodoprivredi broj 2009/1-3.
Analytic hierarchy process: individual and group consistency of decision makers
Essential part of the Analytic Hierarchy Process (AHP) is pair-wise comparison of hierarchy elements and building of corresponding local reciprocal matrices from which weights of the compared elements are calculated. Each matrix, together with the calculated weights, can be used for measuring the decision makers' consistency.
Consistency can be measured in different ways, either directly or indirectly. Standard AHP application provides consistency ratio as a consistency measure, while other methods use the Euclidean distance (ED) as well. If there are K decision makers that evaluate the same problem, then there are also K local matrices at each hierarchy level and K groups of local weights..
Aggregation of weights obtained for each decision in to one (group) decision can be achieved in several ways; and the problem that appears is how to treat the differences in decision makers’ respective consistencies. In this paper we propose an approach how to tackle and solve this problem.
An analytical procedure for identifying decision makers' evaluations that should be included into the group decision is developed and demonstrated by one example. At individual levels, consistency measures used are CR and ED; whereas in group context the Spearman's rank correlation coefficient has been employed. Different 2D and one 3D metrics are used to determine distances of the respective decision makers from ideal points and provide means for implementing different grouping strategies - such as clustering, elimination, etc.
Key words: AHP, decision makers, group, consistency
Bojan SRDJEVIC, Kosana SUVOCAREV, Zorica SRDJEVIC
University of Novi Sad, Faculty of Agriculture, Department of Water Management
E-mails: (bojans, suvocarev, srdjevicz)@polj.ns.ac.yu
Preporučene knjige:
|