Abstract



We make a review of several approaches for ranking alternatives in multicriteria decisionmaking problems when there is imprecision concerning the alternative performances, component utility functions and weights. We assume decision maker?s preferences represented by an additive multiattribute utility function, in which weights can be modeled by value intervals, an ordinal relations, independent normal variables or fuzzy numbers. The performances of the alternatives under consideration are represented by value intervals and classes of utility functions are available for each attribute. The reviewed approaches are based on preference intensity measures or the exploration of the weight space in order to describe the valuations that would make each alternative the preferred one. Three dominance measuring methods are considered, in which a dominance matrix is computed as the starting point, where each element is the minimum of the utility difference between two alternatives. From this matrix, the first approach provides a measure to rank the alternatives. However, the other two methods need to obtain a new matrix denoted as the preference intensity matrix, which provide a measure to rank the alternatives. The difference between both approaches is in how the elements of the preference intensity matrix are computed. On the other hand, two methods based on the exploration of the weight space in order to describe the valuations that would make each alternative the preferred one are considered. Both compute confidence factors describing the reliability of the analysis. The first is the stochastic multicriteria acceptability analysis (SMAA). SMAA computes acceptability indices, which measure the variety of different preferences that give each alternative the best rank. However, SMAA ignores information about the other ranks. This problem is solved with the second method, SMAA2. The performance of the above method is analysed using MonteCarlo simulation on the basis of two measures of efficacy: a) hit ratio, which computes the proportion of all cases in which the method selects the same best alternative; and b) rankorder correlation, which represents how similar the overall structures ranking alternatives are in the true ranking and in the ranking derived from the methods.  
International

No 
Congress

The 21st International Conference on Multiple Criteria Decision Making 

960 
Place

Jyväskylä  Finlandia 
Reviewers

Si 
ISBN/ISSN




Start Date

13/06/2011 
End Date

17/06/2011 
From page

52 
To page

52 

MCDM The 21st International Conference on Multiple Criteria Decision Making 