Computational Approach to Overcome Overlapping of Clusters by Fuzzy k-Means
Katikireddy Srinivas1, K.V.D. Kiran2
1Katikireddy Srinivas, Research Scholar, Department of CSE, KL Deemed to be University, Vaddeswaram (Andhra Pradesh), India.
2Dr. K.V.D. Kiran, Professor, Department of CSE, KL Deemed to be University, Vaddeswaram (Andhra Pradesh), India.
Manuscript received on 15 December 2018 | Revised Manuscript received on 27 December 2018 | Manuscript Published on 24 January 2019 | PP: 350-355 | Volume-7 Issue-4S2 December 2018 | Retrieval Number: Es2079017519/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Of all the clustering algorithms, the frequently employed methods of partitioning algorithms include k-means, medoids and certain modifications. For K-means, a centroid represents the mean or median point of a group and for Kmedoids, wherein a medoid represents the most central point of a data group. We present a hybrid method with both algorithms; k-medoids and k-means to cluster a dataset of thyroid disease drugs and the program is run to generate clusters centred on k-means and k-medoids, followed by enhancing the outcome by implementingfuzzy k-means. Clusterability was carried out by Hopkins statistic and cluster validity by Nbclust resulted in k=3. Both the methods resulted in clusters with negative silhouettes, however, hybrid clustering algorithm resulted in partial overlapping of data points, hence fuzzy k-means algorithm was applied on sub-set of dataset. Finally, of all the six fuzzy algorithms studied, fkm algorithm displayed superior separation of clusters with well-defined data points.
Keywords: k-Means, k-Medoids, Fuzzy k-means, Clustering, Thyroid Disease.
Scope of the Article: Fuzzy Logics