Author(s)

Jiandong Shang, Yunpeng Yang, Runjie Liu, Yaohuan Yang, Panle Li, Huihui Hao

Corresponding Author

Jiandong Shang

ABSTRACT

Taxi plays an important role in the urban passenger transport system. Good taxi service quality will help increase the happiness of urban residents. However, the issue of no-load taxis has always been one of the most concerned issues for urban managers. This paper uses the improved DBSCAN method to carry on the cluster analysis to the taxi trajectory, prompts the taxi driver to avoid the no-load area, and achieves the purpose that to raise the taxi utilization factor. The improved DBSCAN method introduces two distance metrics, segment feature distance and dynamic space-time distortion distance. The feature distances of the segment include the vertical distance, horizontal distance, and angular distance, which can effectively measure the distance between trajectory segments. The dynamic space-time distortion distance introduces spatial factors on the basis of the dynamic distortion distance, and successfully overcomes the similarity error problem caused by different trajectory lengths. At the same time, using the GPS data set of Nanjing taxi to carry out experiments, the results show that the algorithm has higher accuracy and stability than the algorithms in the literature.

KEYWORDS

DBSCAN-Based, no-load road, detection algorithm