Fuzzified Crisp Relative Cure Hierarchical Clustering for Temporal Relational Data
L. Jaya Singh Dhas1, B. Mukunthan2
1L. Jaya Singh Dhas, Research Scholar, Department of Computer Science, Jairams Arts and Science College, Karur, Tamilnadu, India.
2B. Mukunthan, Research Supervisor & Assistant Professor, Department of Computer Science, Jairams Arts and Science College, Karur, Tamilnadu, India.
Manuscript received on 15 August 2019. | Revised Manuscript received on 25 August 2019. | Manuscript published on 30 September 2019. | PP: 1234-1241 | Volume-8 Issue-3 September 2019 | Retrieval Number: C4319098319/19©BEIESP | DOI: 10.35940/ijrte.C4319.098319
Open Access | Ethics and 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: A bitemporal data clustering is a significant solution to the diverse problems for finding the intrinsic structure and compact information over temporal data. The temporal data are collected in the series of particular time periods. The various data mining methods have been developed in the temporal relational data analysis. But the accurate analysis was not performed with minimum time. An efficient technique called Fuzzy Crisp Relative Spherical CURE Hierarchy Clustering (FCRSCHC) is introduced for improving the temporal relational data analysis by partitioning the total dataset into different clusters with minimum time as well as space complexity. The CURE hierarchical structure takes the number of scattered temporal data points in the spherical surface for the clustering. After that, ‘k’ number of clusters and the representative points (i.e. cluster centroid) are initialized. Then the distance between the representative point and the temporal data point are calculated using spherical coordinates. The minimum distance between the data points are grouped into a particular cluster. Then the fuzzy memberships between the two cluster representative points are calculated based on the distance metric. The CURE hierarchical structure merges the two clusters based on the crisp relation between the representative points. Then, the newly obtained clusters are validated using the silhouette coefficient to identify the data points are close to its own cluster or their neighboring clusters. Finally, the optimal numbers of clusters are obtained and minimize the incorrect data clustering which improves the accuracy. The experimental evaluation is performed using a bitmeporal dataset with various parameters such as clustering accuracy, false alarm rate, clustering time and space complexity. The results show that FCRSCHC technique improves the clustering accuracy and minimize the time as well as space complexity as compared to the state-of-the-art- works.
Keywords- Bitemporal data analysis, CURE hierarchical clustering, fuzzy membership, crisp relation, silhouette coefficient
Scope of the Article: Clustering