Title: Clustering trajectory-based objects in spatio networks
Authors: Midde Ranjit Reddy; K.G. Srinivasa; B. Eswara Reddy
Addresses: Department of Computer Science and Engineering, JNTUA, Ananthapuramu, Andhra Pradesh, India ' Department of Information Technology, CBPGEC, Jaffarpur, New Delhi, India ' Department of Computer Science and Engineering, JNTUA, Ananthapuramu, Andhra Pradesh, India
Abstract: Clustering is a proficient approach to breaking down and locate the enormous, concealed, obscure and fascinating information in expansive scale dataset, which encourages the fast improvement of information mining innovation in late decades. With the advancement of area-based administration, moving article clustering turns into a blossoming subject in related fields as a key some portion of information mining innovation. It is a moderately new subfield of information mining which increased high notoriety. This paper considers the issue of proficiently keeping up a clustering of an active arrangement of information focuses that move persistently in 2D Euclidean space. This paper recommends an improved k-means (i-kmeans) algorithm which is done in four stages, which uses segmentation cluster as part of improved k-means. To describe the effectiveness of the obtained cluster we use Silhouette Coefficient metric. Experimental results reveal that improved i-kmeans technique gives better results in terms of accuracy and quality than the traditional one.
Keywords: clustering; k-means; spatial networks; segments; accuracy.
International Journal of Autonomic Computing, 2018 Vol.3 No.1, pp.72 - 85
Received: 04 Dec 2017
Accepted: 13 Mar 2018
Published online: 24 Jun 2018 *