BahasaMelayuMalaysiaEnglish (UK)


Authors: Alvin Ong and Mohd Hanafi Ahmad Hijazi



Global Positioning System (GPS) trajectory prediction is one of the important work in the current era. Clustering on GPS trajectory data will group the similar subjects together and based on the relationship of the cluster, prediction can be made. Due to the nature of trajectory data, we would like to identify the movement patterns of subject in a region. The regional typical movement patterns are also known as profile of moving objects. In this research, we first cluster similar GPS coordinates of a subject into replacement lines. An approach known as DivCluST, that considers both spatial and temporal constraints, is used for the clustering task. Then, the identified cluster is used as features for movement prediction where we can predict the next location the subject may be visited based on the identified movement patterns. These patterns are extracted using Association Rule Mining. Promising results are obtained from the experiments.