Clustering with Intelligent Linexk-Means

Subject Areas : Statistics

نرگس Ahmadzadehgolia ¹ , M.H. Behzadi ² , A. Mohammadpour ³

1 - Department of Statistics, Science and Research Branch, Islamic Azad University, Tehran, Iran
2 - Department of Statistics, Science and Research Branch, Islamic Azad University, Tehran, Iran
3 - Department of Statistics, Faculty of Mathematics & Computer Science, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran.

Received: 2017-12-19 Accepted : 2018-07-04 Published : 2018-08-01

Keywords: معیار عدم تشابه, خوشه‌بندی, k-میانگین هوشمند, تابع زیان لاینکس,

Abstract :

The intelligent LINEX k-means clustering is a generalization of the k-means clustering so that the number of clusters and their related centroid can be determined while the LINEX loss function is considered as the dissimilarity measure. Therefore, the selection of the centers in each cluster is not randomly. Choosing the LINEX dissimilarity measure helps the researcher to overestimate or underestimate the centers which helps to assign some entities into a special cluster. We check the performance of the algorithm on some real and artificial datasets and evaluate the results according to some internal and external indexes.