A modified algorithm to find a representative capacity with evenness consideration for non-additive robust ordinal regression

Non-additive robust ordinal regression (NAROR), uses primary preferences of decision maker (DM) to define the necessary preference relations (NPR) and possible preference relations (PPR) on alternatives when all compatible fuzzy measures are taken into account and aggregation function is Choquet integral (CI). The question arises as how these NPRs and PPRs can be used in capacity definition problem? This article proposes an algorithm which uses these relations in finding a capacity that is representative to DM and has also evenness property to some extent. The methods based on maximizing evenness leads to results that are not fully representative to DM, so this article improves this drawback by focusing on representativeness of capacity and taking into account capacity evenness by means of decision rules defined in algorithm.