Statistical analysis of MCDM data normalization methods using Monte Carlo approach. The case of ternary estimates matrix
Date
2018Author
Kosareva, Natalja
Krylovas, Aleksandras
Zavadskas, Edmundas Kazimieras
Metadata
Show full item recordAbstract
One of the most important stages of solving MCDM problems is normalization of initial decision-making matrix. The impact of 5 widely used normalization methods on the best alternative determination accuracy in the case of ternary estimates decision matrix is analysed in the article. Alternatives ranked by applying SAW method. Monte Carlo procedure was conducted fordata matrices of different dimensions and both optimization directions. Two cases - the more and the less separable alternatives- were analysed.None of the 5 methods were the best or the worst in all cases. Nevertheless, Minmax method inmost cases is significantly better than other. The Log method is the worst in some cases, but it is the best (or one of the best) in other cases.The highest values of the best alternative detection accuracy were accompanied by the lowest standard deviations of experiment results, respectively, the lowest values – by the highest standard deviations.