Integrating Indirect Preferences Into Ordinal Regression Estimation of Missing Values and Consensus Reaching

Constraints from the external environment and personal experiences may hinder decision makers from offering complete fuzzy preference relations in situations of group decision making. Conflicts in individual different preferences impede any effort of reaching consensus. Addressing such challenges, traditional methods typically isolate fuzzy preference relation completion and consensus reaching as two separate parts, while improving consistency level through iterative processes. However, such an approach affects the efficiency of consensus achievement. Furthermore, existing studies utilize limited known objective information for filling missing values, while neglecting the excavation of indirect preferences that requires lower cognitive capacity from decision makers. Consequently, the credibility of decision results is greatly reduced. Through integrating indirect preferences, this article introduces a consensus framework that concurrently accomplishes fuzzy preference relation completion and consensus reaching. In the framework, an ordinal regression consensus discriminant model is constructed to first assess the compatibility of information sets. Subsequently, to accurately identify and manage conflicting information, we design a conflict feedback mechanism by considering the minimum cost elimination consensus model and the minimum cost adjustment consensus model. Our proposed method not only deeply explores decision makers’ implicit preferences, but also achieves the integration of three goals: missing value prediction, consensus reaching, and additive consistency realization. It avoids multiple consistency iterations and distortions, while enhancing the reliability of estimated values and the efficiency of consensus reaching. Finally, the validity of the proposed method is demonstrated through case analysis, sensitivity analysis, and comparative analysis.