Similarity based fuzzy TOPSIS with OWA: Reflecting risk attitudes in multiple-criteria and multi-expert evaluation under uncertainty

We present similarity based fuzzy Technique for Order of Preferences by Similarity to Ideal Solution (fTOPSIS) using ordered weighted averaging (OWA) operators. With OWA operators we extend the use of similarity based fTOPSIS in two ways. First in several types of real world engineering decision making problems the amount of criteria required to be satisfied to a high extent is more important than satisfying particular criteria; also in the aggregation of expert evaluations the amount of experts providing good/bad evaluations might be critical. This kind of generalization is now made possible by using OWA in the proposed method. Second by using linguistic quantifiers to define weights for OWA and allowing different quantification for the similarity to Fuzzy Positive Ideal Solution (FPIS) and Fuzzy Negative Ideal Solution (FNIS) we allow decision-maker’s risk preference to be reflected. This allows us to model risk-averse or risk-seeking decision making in fTOPSIS for the first time. Aggregation of multiple expert evaluations using fTOPSIS that considers specific linguistically expressed risk-attitude of the user of the final evaluation is now also enabled in fTOPSIS. The advantage of the proposed method lies in the ability to adjust the method according to decision makers’ risk preferences. Multiple expert evaluations can also now be aggregated according to the end-user’s risk preference. The proposal enables separate treatment of FPIS and FNIS using OWA operators to reflect decision maker’s optimism/pessimism. Presented findings show that by doing this we are able to reflect risk attitudes and receive different ranking corresponding with risk preferences. • Similarity based fuzzy TOPSIS with linguistically induced OWA weighting is proposed. • Linguistic quantifiers induce two sets of OWA weights to reflect risk attitude. • OWA weights allow for asymmetric treatment of ‘similar to PIS’ and ‘similar to NIS’. • The method is applicable in single and multiple-expert evaluation under uncertainty. • Aggregation of expert evaluations obtained through various methods is made possible.