TSX-Means: An Optimal K Search Approach for Time Series Clustering

Proliferation of temporal data in many domains has generated considerable interest the analysis and use time series. In that context, clustering is one most popular mining methods. Whilst series algorithms generally succeed capturing differences shapes, they often fail to perform based on both shape amplitude dissimilarities. this paper, we propose a new method automatically determines an optimal number clusters. Cluster refinement dispersion criterion applied distances between their representative within cluster. That measure allows for considering We test our datasets compare results with those from K-means (TSK-means) K-shape