Mining algorithm of spatial-temporal co-occurrence pattern based on vehicle GPS trajectory

Abstract

In the calculation process of spatial-temporal co-occurrence patterns, traditional methods often set the whole time frame as the actual existence time by default for all moving targets. However, in practice, existence time frame of different types is not necessarily the whole time frame. Based on this fact, the paper describes the calculation method of spatial-temporal interest degree-spatial frequency and time frequency in order to improve the practicability of co-occurrence patterns. In addition, the paper sets spatial-temporal weight coefficient for every pattern and sorts all candidates of co-occurrence patterns based on their weight. Then high efficiency co-occurrence patterns can be selected easily. Thus the proposed algorithm in this paper provides a solution to the difficulty of setting time thresholds and space thresholds in advance. And the experiment results show that the method can improve the effectiveness of spatial-temporal co-occurrence patterns simultaneously.