Different Representations Ensemble with Temporal Data Clustering Via Weighted Clustering

1Vamsi Krishna Jayanti and 2Vasantha Kumar

1Department of Information Technology, Email:,

2Department of Computer Science and Engineering, Avanthi Institute oF Technology, Makavanipalem, Narsipatnam, Andhra Pradesh, India.





Temporal data clustering provides underpinning techniques for discovering the intrinsic structure and condensing information over temporal data.  In this paper, we present a temporal data clustering framework via a weighted clustering ensemble of multiple partitions produced by initial clustering analysis on different temporal data representations. In our approach, we propose a novel weighted  consensus function guided  by clustering  validation criteria to reconcile  initial partitions  to candidate consensus partitions  from different perspectives, and  then,  introduce an agreement function to further reconcile  those candidate consensus partitions to a final partition. As a result, the proposed weighted clustering  ensemble algorithm provides  an effective enabling  technique for the joint use  of different representations, which cuts the information loss in a single representation and exploits various information sources underlying  temporal data.  In addition, our approach tends to capture the intrinsic structure of a data  set,  e.g.,  the number of clusters. Our approach has  been evaluated with benchmark time series, motion trajectory,  and  time-series data  stream clustering  tasks. Simulation results demonstrate that our approach yields favorite results fora variety of temporal data  clustering  tasks. As our weighted cluster ensemble algorithm can combine any input partitions  to generate a clustering  ensemble, we also  investigate its limitation by formal analysis and empirical studies.

    Index Terms—Temporal data clustering, clustering ensemble, different representations, weighted consensus function, model selection.




International eJournal of Mathematical Sciences, Technology and Humanities

Volume 2, Issue 5, Pages:  830 - 837