Research on the clustering algorithm of the bicycle stations based on OPTICS

With the development of the sharing traffic economy, the urban fixed‐station parking public bicycle system greatly facilitates the residents to travel. However, the tidal problem has greatly reduced the social utility of the public bicycle system. A step‐by‐step scheduling strategy has been put forward to solve the problem, which firstly clusters the stations according to station locations or the number of bicycles circulating in a station and then perform regional scheduling in each cluster. However, because the clusters obtained by this method do not have the characteristics of a closed small‐world network, the strategy does not cover all stations that require bicycle scheduling. To fix the gap, a station clustering algorithm is proposed, which employs SimRank to calculate the similarity between the stations according to the loan‐to‐return relationship and then adopts the density clustering algorithm of OPTICS (ordering points to identify the clustering structure) to cluster all stations based on the similarity. The data of public bicycle system in Ningbo City, Zhejiang Province of China, is employed to validate the algorithm. The experiment found that the station clustering results have distinct regional characteristics, and the algorithm is competent to divide the scheduling area of public bicycles.