A Parallel Algorithm for Mining Time Relaxed Gradual Clustering Pattern Based on Spatio-Temporal Trajectories
Hongyan Sun, Genlin Ji, Bin Zhao, Xintao Liu · 2017
As an important area of spatio-temporal data mining, time relaxed gradual clustering pattern has been attracting broad attention in recent years. Algorithm PTRGSP is proposed to mining time relaxed gradual clustering pattern from spatio-temporal trajectory, which is implemented under spark framework. All trajectories are divided into point sets with different timestamps, and such point sets are parallel clustered. After that, clusters of different time intervals are joined in parallel to achieve pattern candidates. Candidates are combined to obtain interesting maximal time relaxed gradual clustering pattern. To improve the efficiency of the algorithm PTRGSP, algorithm PTRGSP-G is presented based on grid index for mining time relaxed gradual clustering pattern. The experiment results on real dataset and synthetic trajectory dataset demonstrate the effectiveness and efficiency of the two algorithms.