Cluster Analysis of Stations Based on Weight SimRank in Sharing Bicycle

Main Article Content

Bo Guan, et al.

Abstract

With the increasing popularity of shared bikes, the indiscriminate parking of bicycles in cities has increasingly become a difficulty in urban management. The tidal phenomenon of large numbers of urban residents during rush hour is the root cause of the indiscriminate parking of bicycles in many subway stations and commercial areas. Optimizing the scheduling strategy of shared bikes is one of the effective solutions to solve the problem of random parking and reduce the scheduling cost. The cycle is a short - distance vehicle, and its circulation law is in line with the characteristics of the small world of urban traffic. That is, most of the bikes flow within the small world region, while only a small part of the bikes flow between the small world regions. With the massive accumulation of bike-sharing borrow and return data, the method of clustering the borrow and return stations and dividing regions according to the clustering results has attracted the attention of industry experts and researchers. it is effectively to apply in intelligent scheduling related industries. Although there have been some studies on the station clustering in the current literature, because these studies are basically based on the fixed features of the site (site location, pile number, etc.), the results cannot find an effective small world region of bicycles. In order to find out the effective small world region of bicycles, we introduced the idea of SimRank (that is, the similarity of a station is due to the similarity of its bicycle source station and destination station), and assigned weights to association relationships (the number of times of borrowing and returning) to define the similarity algorithm w-SimRank of stations. Then, the station clustering was done in line with skyline thinking. Finally, in order to verify the effectiveness of the algorithm, we implemented the station clustering based on SimRank algorithm, and compared the clustering effect with the W-SimRank algorithm proposed in this paper to verify the effectiveness of the W-SimRank algorithm, and analyzed the influence of the key parameters of the algorithm on the algorithm. And then

Article Details

How to Cite
et al., B. G. (2021). Cluster Analysis of Stations Based on Weight SimRank in Sharing Bicycle. CONVERTER, 09 - 18. https://doi.org/10.17762/converter.152
Section
Articles