Year: 2021
Type: conference-paper
Abstract: It is common practice in the MCDA to use several multi-criteria decision methods and then compares obtained rankings with one or two different rank correlation coefficients. The problem is that different rank correlation coefficient gives different values for the same pair of rankings, and the number of studies which tries to investigate it is small. Studying the similarity of rankings is a very important challenge in multi-criteria decision support, and the coefficients themselves seem to be the most practical ways of evaluating rankings. This paper compares chosen rank correlation coefficients to show how much different they are. Spearman’s, Weighted Spearman’s, Kendall Tau and Rank similarity correlation coefficient are compared statistically. The paper confirms that the coefficients are closely related, and their dependence is graphically represented, which initiates research towards allows for their better selection in the future. In conclusions, directions of further development are indicated.
Source: Procedia Computer Science
Author: Andrii Shekhovtsov
Institution: West Pomeranian University of Technology in Szczecin
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: diamond
Award: G8185403414
OpenAlex ID: W3203448770