Where You Really Are: User Trip Based City Functional Zone Ascertainment

As one convenient and green transportation mode, Public Bike System (PBS) offers increasing service. It has generated a great mass of transit records as a famous application around the world. To mine the trip information in the records is an effective way to understand the city from some aspects such as the city functional zone distribution and the crowd mobility. According to the observation from daily life, it's quite often found that some users are willing to shift their bikes at some intermedia stations so as to decrease charge. The phenomenon indicates that the drop-off stations in the system records may not be their real destinations. Focusing on this phenomenon, this paper proposes a new method named User trip based city functional Zone Ascertainment (UZA) to redefine the city functional zone. This paper divides the city into adjacent and non-overlapping regions by the Voronoi Diagram method, and discovers the topic feature distribution of each region based on the user trips. By clustering, we then integrate regions into zones and finally ascertain their functions. Extensive experiments are conducted on the real dataset collected from Hangzhou PBS, consisting of more than 23 million transit records. Compared with the Hangzhou city planning graph, the result indicates our method is reliable and effective.