Data segmentation through two-level clustering with greedy approach

¹ Reshetnev Siberian State University of Science and Technology, 31, Krasnoyarskii rabochii prospekt, Krasnoyarsk 660037, Russian Federation
² Siberian Federal University, 79 Svobodny Prospekt, Krasnoyarsk 660041, Russian Federation

Abstract

This study presents a two-level clustering method utilizing a simplified greedy procedure to enhance data processing efficiency and accuracy, particularly with large high-dimensional datasets. The two-level structure allows for the identification of broad data groups in the first stage, followed by a more granular analysis within these groups in the second stage, thereby accelerating the clustering process and improving result quality. The application of the k-means++ method did not yield the anticipated benefits compared to traditional random initialization. Such findings underscore the necessity for preliminary data analysis when selecting optimal clustering algorithms, as instances of complex methods failing to improve results are not uncommon. This work illustrates the importance of balance between method complexity and effectiveness in real-world applications and emphasizes the potential for increased resource expenditure without commensurate gains in clustering performance.