International Academic Journal of Science and Engineering

  • ISSN 2454-3896

A New Approach to Trajectory Data Clustering

Anahita Ghalami and Hassan Naderi

Abstract: Data analysis of the moving object trajectories and extracting their moving patterns using data mining methods has attracted the interest of researchers in recent years. In many moving objects management programs, analysis of data clustering has turned out to be an essential requirement. Clustering is one of the most significant methods of analysis that groups similar data to produce a summary of the data distribution patterns in a dataset. In this paper, after examining the existing algorithms for clustering the moving object trajectories in an environment where moving objects are not limited in road networks, one of the most prominent algorithms in the field called Traclus was selected. The algorithm rests on the partition-and-group framework. In this algorithm for clustering subtrajectories, only the latitude and longitude points of the trajectory are taken into account. Therefore, this algorithm can only identify dense points without considering the time when these areas have been dense. However, in many cases, it is essential to determine what areas have been dense and at what times. Therefore, with considering the temporal characteristics of the trajectory points, the algorithm was developed and our own algorithm called T-Traclus was proposed. The advantage of this method over existing methods is its capability of detecting dense areas at the same time

Keywords: Spatio-temporal databases, moving objects trajectory, clustering, trajectory clustering

Page: 41-52

Volume 3, Issue 1, 2016