Mining Frequent Trajectory Patterns from GPS Tracks

Abstract

As recent advances and wide usage of mobile devices with positioning capabilities, trajectory database that captures the historical movements of populations of moving objects becomes important. Given such a database that contains many taxi trajectories, we study a new problem of discovering frequent sequential patterns. The proposed method comprises two phases. First, we cluster the stay points of taxis to get collocation patterns for passengers. Then, for each pattern instance, we use an efficient graph-based searching algorithm to mine the frequent trajectory patterns, which utilizes the adjacency property to reduce the search space. The performance evaluation demonstrates that our method outperforms the Apriori-based and PrefixSpan-based methods.