High Utility Co-location Patterns Mining from Spatial Dataset with Time Interval

A spatial co-location is a set of spatial events being frequently observed together in nearby geographic space. The existing high utility co-location patterns mining algorithm are based on idealized data, without considering time interval of spatial instance in real-world. In order to address the problem, firstly, this paper redefines the concepts of spatial instance, the minimum time interval overlap rate and neighbor relationship, etc. Secondly, a basic algorithm and a pruning algorithm of high utility co-location patterns mining with time interval are proposed. Since high utility co-location patterns mining do not satisfy the anti-monotonic property, a pruning strategy based on feature utility participation weight is exploited for computational efficiency. Finally, experimental results are provided to show the effective, efficient, and scalability of the basic and pruning algorithm. In addition, the algorithm proposed in this paper is also compared with high utility patterns mining algorithm without time interval.