RANCOM: A novel approach to identifying criteria relevance based on inaccuracy expert judgments.

The identification of the criteria relevance from the expert in the multi-criteria problem could be a challenging task. It is important to provide simple and intuitive techniques that reflect the expert knowledge effectively. When the problem includes many criteria, it is possible to occur some probability of error in determining the criteria relevance. However, most subjective weighting methods perform correctly when the experts do not hesitate, are repeatable, and are very precise. Consequently, there is interest in improving the currently used approaches to handle expert hesitation more effectively. To fill this gap, we propose the RANking COMparison (RANCOM) method, which can be used to determine criteria weights based on expert knowledge. Its performance is based on defining the criteria ranking order to obtain the weights vector. The illustrative example and a comparative analysis of the selected subjective weighting methods show that the proposed approach provides reliable results, highly correlated with the existing approaches. However, it offers a simple manner of use, adjusted for less and more experienced users. The comparison of the RANCOM and AHP methods shows that when a slight probability of error in expert judgments occurs, the RANCOM method proves to be a more suitable solution to handle the expert inaccuracies for the problems with 5 or more criteria. Performed experiments showed that the RANCOM method is a repeatable and more consistent method regarding possible judgment errors. The obtained results showed that the RANCOM method guarantees more repeatable results than the AHP method, especially for a greater number of criteria. [ABSTRACT FROM AUTHOR]