Study Of The Combination Of Multi-attribute Decision Making Method

As one of the most important research areas in the science ofdecision-making, Multi-attribute decision-making has been used invarious aspects of socio-economic life. However, because of thedifferences of the decision problems, the complexities of availabledecision information and the diversities of the feasible decision-makingmethods, the appraisal results are short of stabilty and reliability, and thedecision-making will be affected. Consequently, studying the methodsand measures to improve the stability and reliability of the appraisalresults and make the evaluation results closer to the real results has greatacademic value. At the same time, put these methods and measures inpractice can reduce the situations of ambiguity and help the managers tomake the decisions, so its practical application value is great too.Multi-attribute combinational decision-making is one of the mosteffective ways to improve the stability and reliability of the evaluationresults. It is a measure that considers the background of thedecison-making problems,uses every kind of decison-making informationcoming from subjective or objective and applies more than one method ormodel to debase the differences, while at the same time supplement witheach other's advantages mutually. This paper employs the thinking ofanalysis and synthesis, studies the problems of combinational weighting,the problems of evaluating and choosing the decision-making methods,the problems of information aggregation and the problems of systematicalmulti-attribute combinational decision-making methods and theirsapplication. During the different phases of the process of decision-making,a series of models that can meet the decision-making practical need areproposed. A lot of examples are given to show the feasibility andavailability of those models. The main works in this paper are as follows:(1) Research on subjective and objective integrated approaches todetermine attribute weights under uncertain environment. When attributevalues are certain while subjective weights are uncertain, we can optimizelocally first, then fix on the best concordant subjective weights by themethod of maximizing deviations, and then combines subjective weights and objective weights by using the least squares method. When subjectiveweights are certain while attribute values are intervals numbers, triangularfuzzy numbers, trapezoidal fuzzy numbers, we define the distancebetween two fuzzy nubers first, and then extend combinational weightingmodels from certain environment to uncertain environment. Acombinational model (to deal with this kind of problems) based on thedegree of fuzzy deviation is proposed, and an illustrative example withinterval numbers is given to demonstrate its effectiveness.(2) Study on how to evluate and choose methods to constructcombinational methods set. To concrete combinational decision-makingproblems, this paper follows the principle of roughly-chosen first andwell-chosen later. The first step is to choose feasible methods, and thenchoose methods from feasible methods set to construct combinationalmethods set. Theoretically, it is difficulty to prove one decision-makingmethod is more applicable than the other, so this paper studies themethod-chosen problem quantificationally on the basis of independentappraisal results of the single methods. Four models to choose applicablemethods that based on similar coefficients, deviation, degree ofcompatibility and discrepancy, fuzzy clustering in fuzzy mathematics areconstructed respectively. These models estimate the closeness from onesingle method's appraisal results to the others in the different aspacts, andregard this as the basis of the method-chosen. Four models are employedto choose the evaluation methods in the example of evaluating thefinancial situation among the listed companies, the consistency of theresults reveal these models' excellent properties.(3) Research on the problems of synthesizing several independentappraisal results in multi-attribute combinational decision-making. whenthe weights of the methods are given, a combinational model based on theidea of fuzzy clustering is proposed. In this model, to get thecombinational result just similar to the process of getting the clustercenter in fuzzy c-means clustering. The paper provides the expression toget the result and it is proved by apagoge. In the situation that the weightsof the methods are not given, the appraisal results of a single methodbecome a random variable for theirs discrepancy among the different independent methods, and the appraisal results under each single methodwhich is employed in the real problem can be interpreted as a sample.Therefor, the weight of each method can be interpreted as the probabilitywhen the true result chooses the corresponding sample. Apparently, theweights of the methods have uncertainty. For this reason, this paperconsider the weights' uncertainty of the methods on one hand, and at thesame time, think over the deviation between the combinational resultsand each single appraisal results for the other, synthetically use theprinciple of the greatest Jaynes entropy and optimization theory, proposeda appraisal results combinational model based on entropy under thesituation that the weights of the methods are not known. In addition, thispaper applies a famous operator in information aggregation area namedOWA to combine independent appraisal results, and the model based onthe OWA operator is proposed, the instance verifies its feasibility.(4) Integrates all the phases in the process of decision-making, thesystematical multi-attribute combinational decision-making method andits application are studied from the level of methodology. The basicconnotation of the systematical multi-attribute combinationaldecision-making is defined and the systematic framework is discussedfrom the aspacts of procedure and computer system framework. At last,the paper gives an example to illustrate the procedure when thesystematical multi-attribute combinational decision-making method isemployed in an evaluation problem for a lake-district's agriculturalindustrialization.