基于改进DBSCAN算法的出租车载客热点区域挖掘研究

Rearch on Mining Taxi Pick-up Hotspots Area

出租车载客热点区域对出租车寻客具有指导意义,针对现有密度聚类方法存在聚类精细程度不足的问题,提出改进的DBSCAN密度聚类算法用于出租车载客热点区域挖掘.首先,以道路交叉口、道路转弯点为节点提取道路拓扑结构;然后,使用A*算法进行寻路并以路段为单位提取邻域内载客点进行聚类,加快聚类速度.最后,使用成都市浮动车订单轨迹数据与订单数据进行验证,与传统的DBSCAN算法相比,本算法可以挖掘出更精细的载客热点,同时具有更好地稳定性.

The taxi pick-up hotspot area has guiding significance for taxis riders. Aiming at the problem that the existing density clustering method has insufficient clustering degree, this paper proposes an improved DBSCAN method for mining taxi pick-up hotspot areas. First, the topology is obtained by taking road intersections and road inflection points as nodes;Then, this paper uses the A* algorithm and passenger points to extract and cluster road segments. This paper uses Chengdu s floating car trajectory data and order data provided by DiDiChuXing for verification. Compared with DBSCAN, this algorithm can mine more detailed taxi pick-up hotspot areas and has better stability.