ABSTRACT
Clustering, as an important area of data mining, solves the problem of partitioning a set of objects into homogeneous groups based on the similarity of the analyzed features. Currently, despite the large number of clustering methods, there are no universal ones. The result depends mostly on the structure of the researched data as well as the ways of formalizing the knowledge of objects and clusters similarity. It is important to evaluate the quality of the results in order to select the partition which is most similar to the cluster structure relevant to the researched data. For this reason, the article presents the information technology of multi-criteria quality evaluation and increase in clustering results sustainability on the basis of the relative quality criteria, decision making theory methods and a set of algorithms.
