Learning taxis’ cruising patterns with Ripley’s K function

ZONG Fang(宗芳)¹, ZHANG Hui-yong(张慧永)², LI Hai-fan³

(1. College of Transportation, Jilin University, Changchun 130022, China;
2. School of Management, Jilin University, Changchun 130022, China;
3. College of Engineering, Georgia Institute of Technology, GA 30332, USA)

Abstract:Taxi drivers’ cruising patterns are learnt with GPS trajectory data collected in Shenzhen, China. By employing Ripley’s K function, the impacts of land use and pick-up experience on taxis’ cruising behavior are investigated concerning about both intensity of influence and radius of influence. The results indicate that, in general, taxi drivers tend to learn more from land use characteristics than from pick-up experience. The optimal radius of influence of land use points and previous pick-up points is 14.18 km and 9.93 km, respectively. The findings also show that the high-earning drivers or thorough drivers pay more attention to land use characteristics and tend to cruise in high-density area, while the low-earning drivers or focus drivers prefer to learn more from previous pick-up experience and select the area which is far away from the high-density area. These findings facilitate the development of measures of managing taxi’s travel behavior by providing useful insights into taxis’ cruising patterns. The results also provide useful advice for taxi drivers to make efficient cruising decision, which will contribute to the improvement of cruising efficiency and the reduction of negative effects.

Key words:taxi; cruising patterns; land use; pick-up points