Böyük həcmli məlumatların klasterləşdirilməsi üçün DBSCAN alqoritminin modifikasiyası

Abstract

The development of Information and Communication Technologies (ICT) has led to the rapid growth of digital information and the consequent emergence of the concept of large-scale data. Therefore, there is a need to delve into large-scale data and its essence, the possibilities and problems of analytical technologies. Clustering is one of the main methods of analyzing big data. The main purpose of clustering is to separate data into clusters according to certain characteristics. When clusters come in different sizes, densities, and shapes, the problem of detection arises. The article explores the density-based DBSCAN clustering algorithm for working with big data. One of the main features of this algorithm is to create an effective cluster by detecting the noise points in big data. During the implementation of the algorithm, real databases containing noise points were used. Metrics such as adjusted rand index, homogeneity, Davis-Boldin index were used to evaluate the results of the experiment. The proposed method was more effective than the traditional DBSCAN algorithm in detecting noise points.