Research On Uncertain Multiple Attribute Decision Making With Preference Information

Multiple attribute decision making, which is an important part of modern decision science, is to choose the best alternative and rank the alternatives under many attributes. Generally speaking, because of the complexity of objective things and the uncertainty of the human mind involved in real-world decision problems, there are lots of uncertain multi-attribute decision-making problems. Research on this kind of problems has received wide attention of scholars all over the world, but it remains to be further investigated in the solving method. This thesis studies the uncertain multi-attribute decision-making problems with a variety of ways and types preference information at the same time.Decision-makers give preference information in different ways, which can be on attributes preference information or alternatives. The preference information is referred to subjective information of the decision problem. Due to the different cultural background and knowledge structure of decision-makers, the types of preference information are different, such as, ordinal interval, language variables, interval number and fuzzy subset. However, in the decision analysis, if the subjective information is only considered, the decision results are subjective. And if the decision matrix is considered, which includes objective information, the results will lose practical significance. Therefore, the objective and subjective information needed to be simultaneously considered so that the results will be more effective. According to the preference information in different ways, the thesis proposes the solution of multiple attribute decision making problems, which is based on integration of subjective and objective information in different ways.Firstly, preference information on attributes is given by decision-makers, which can be contained the interval number and relational inequality. The thesis proposes a new modeling method that makes use of subjective and objective information to determine attribute weights, by combining these two types of preference information. In addition, a corresponding example is given to verify the feasibility of the method.Secondly, preference information on alternatives is given by decision-makers are in different types. A model was set up by combining the objective decision matrix and the subjective complementary matrix derived from the preference information. Besides, a corresponding example is given, and solved by Matlab, verifying the feasibility of the method.Finally, because of the different thinking angles of decision-makers, they may provide preference information on attributes and alternatives in a decision problem. From this case, the thesis proposes a comprehensive approach from a new perspective, which establishes a new optimization model, including decision matrix, attributes preference information, and alternatives preference information. Besides, by giving the corresponding examples, and solving it by Matlab, the thesis gets the final ranking of alternatives, which shows the effectiveness and feasibility of the method.