Multicriteria decision analysis for structural decision problems

Abstract

Generally, a decision problem can be modeled as one of the three types: structural, semi-structural, and non-structural. The structural problems reveal a well organized formulation such as the high structured information tables, the measurable decision goals, and the clear boundary of problems. Both semi-structural and non-structural problems are mostly at the strategic management level. In these problems, decision information is usually non-quantitative or a lack of organization. More importantly, the human perception and judgments do play a decisive role whereas such human factor is usually intangible or unmeasurable. This thesis dedicates to the investigation of Multicriteria Decision Analysis for solving the structural decision problems. The thesis proposes an in-depth research on Multicriteria Decision Making from three aspects: (1) decision theories/methods related to dominance-based rough sets; (2) decision models for multicriteria ranking, sorting, and choice; and (3) decision support systems towards group and uncertain decision-makings. In the theory/method aspect, a new development on dominance-based rough methodology is proposed from the theoretical aspect. We developed a new Dominance-based decision Rule Induction mechanism (DRI for short). Unlike the previous rule induction methods that neglect the values of uncertain information within rough boundary regions, the proposed rule induction method can utilize both certain and uncertain decision information. This method is proved to be more sufficient and more effective than the previous methods, and applied for construction of decision models for multicriteria ranking and sorting. In the model aspects, four new decision models are proposed towards the practical decision problems including supplier selection, warehouse evaluation, and personnel evaluation. First, we developed a novel believable rough set approach model (BRSA model for short) for multicriteria sorting problem and applied this model in supplier selection. Second, we developed a novel Interval Valued Intuitionistic Fuzzy Group Decision model for warehouse evaluation (IVIFGD model for short). This work led to a novel rule-based solution for complex warehouse evaluation problem under interval valued intuitionistic fuzzy environment. Third, we developed a novel Intuitionistic Fuzzy Superiority and Inferiority Ranking decision model (IFSIR model for short) for supplier selection under intuitionistic fuzzy environments. By using the proposed model, the inherent uncertainty in the form of intuitionistic information can be well propagated in the multi-step decision process. Fourth, we developed a novel dynamic tolerant skyline decision model for personnel evaluation (T-skyline model for short). This work additionally provided a detailed empirical study on NBA player evaluation in 2011-2012 regular seasons and consequently revealed several very interesting and valuable evaluation results with the realistic significance. In the system aspect, two designs of decision support system (DSS) are proposed towards group decision-making and uncertain decision-making, respectively. We firstly proposed a framework of ON-TOlogy-based Group Decision Support System (ONTOGDSS) for decision process which exhibits the complex structure of decision group. Secondly, we developed a framework of Uncertainty-based Group Decision Support System (UGDSS). It provides a platform for multiple processes of decision analysis in six aspects including decision environment, decision problem, decision group, decision conflict, decision schemes and group negotiation. As a summary, this PhD research contributed a comprehensive investigation towards solving structural decision problems from multiple perspectives including the decision mathematical tool Dominance-based rough sets (the theory aspect), the realistic problem-solving decision models (the model aspect), and the effective decision support systems (the system aspect).