Statistical analysis of KEMIRA type weights balancing methods

Abstract

The article analyzes Multiple Criteria Decision Making (MCDM) problem when there are two different groups of evaluating criteria. It was shown how criteria weights can be calculated according to weights balancing method by formulating optimization task. Case study of the small dimensions problem was solved by Kemeny Median Indicator Ranks Accordance (KEMIRA) method with options re-selection. Next, 8 various candidates sorting algorithms - 6 based on voting theory methods and 2 algorithms based on Kemeny median - were compared with each other. Monte Carlo experiments were conducted for the cases of 3-10 experts, 3-5 candidates and probability values of correct decision p=0.4-0.8. The highest percent of correct decisions and the lowest percent of failed voting procedures were demonstrated by algorithms based on Kemeny median.