Mojtaba Jahanian - Department of Computer Engineering, Faculty of Engineering, Arak Branch, Islamic Azad, University, IRAN
Abbas Karimi - Department of Computer Engineering, Faculty of Engineering, Arak Branch, Islamic Azad, University, IRAN
Faraneh Zarafshan - Department of Computer Engineering, Faculty of Engineering, Arak Branch, Islamic Azad, University, IRAN

Clustering is one of the most important machines for learning unseen algorithms. Data is not labeled inclustering. The starting point of clustering is very important. Choosing the right number of spikes forclustering is like choosing the right seed. The performance of the algorithm depends on selecting theappropriate number of clusters and selecting the optimal centers. Cluster quality and optimal number ofclusters are important in cluster analysis. In this article, we have tried to differentiate our work fromother existing articles by careful analysis and comparison of existing algorithms, and a clear andaccurate understanding of all aspects. Research compared to other articles Due to the importance ofselecting the number of clusters, we present a smart algorithm in this paper (SONSC: Select the OptimalNumber of Smart Clustering). The proposed SONSC algorithm provides an index that can performclustering with higher accuracy by considering the three criteria of minimum internal distance betweenpoints of a cluster and maximum external distance between clusters and considering cluster level(number of clusters).

Clustering algorithms, K-means, clustering, optimal number of clusters