Özet:

In this study, to present an integrated approach to clustering analysis based on multi-objective decision making, it is aimed to determine the best clustering algorithm among 11 different clustering algorithms by evaluating all 27 internal validity criteria simultaneously with MULTIMOORA method. In the study, initially, the best clustering algorithm was determined according to the optimal number of clusters for two COVID-19 datasets. Then, it focuses on determining the relationship of the country clusters with the classes determined according to the human development index. In the result of the analyses, countries affected by the COVID-19 pandemic have clustered via the CLARA and SOM algorithms according to their proximity calculated from the Euclidean distance. Three optimal number of clusters were determined for both datasets. The incidence rate variable is the more dominant factor than case fatality rate in the real difference between clusters. Another remarkable finding is that while countries with economic power and a high level of human development are expected to be less affected by the pandemic before the vaccination, the level of being affected by the pandemic increases in terms of both variables as the level of human development increases.

Anahtar Kelimeler: