It is known that the quality of clustering depends on the structure of the data and the methods used in the clustering. Clustering methods produce different results depending on the size and structure of the data. Therefore, ensemble methods have been used lately to ensure efficiency, quality and stability of clustering. One of the most difficult problems at this time is the establishment of a useful function for assessing the quality of clustering.

Determination of the utility function depends directly on the abundance of data, as well as the selection of ensemble techniques and their weight in the ensemble. The main contribution of work is as follows: 1) if the class of each object in the data set is known in advance, then Purity as a function of utility;2) if the class of objects in the data set is not known in advance, then the Davis-Bouldin index is selected as a utility function. One of the main advantages of the method is that the weight of the methods included in the ensemble is determined by consensus without expert intervention.

In the article "Weighted consensus clustering and its application to Big data" (doi.org/10.1016/j.eswa.2020.113294)  method was evaluated from various aspects. The experimental results show the effectiveness of the weighted consensus clustering method in comparison to methods included in the ensemble.

The article was published in  “Expert Systems with Applications”  journal. The Impact Factor of the journal is 4.292, is a Q1 class based on both Web of Science and Scopus

This work was financed by the Science Development Fund under the President of the Republic of Azerbaijan.  (Grant № EİF-KETPL-2-2015-1(25)-56/05/1)

The article co-authored by Vice-President of ANAS, Director of the Institute of Information Technology, Academician Rasim Alguliyev, Head of Department, Corresponding Member of ANAS, Doctor of Technical Sciences Ramiz Aliguliyev and senior researcher, Ph.D. Lyudmila Sukhostat.

© All rights reserved. Citing to www.ict.az is necessary upon using news