SVD-DBSCAN-Based Demand-Responsive Feeder Transit Demand Mining at Hub Station

To obtain the spatio-temporal distribution characteristics of demand-responsive feeder transit at hub stations to improve the feeder capacity and attractiveness of public transportation, this paper conducts demand mining for demand-responsive feeder transit at urban hub stations based on cab GPS data. The spatio-temporal matrix of transportation demand considering the constraints of hubs is obtained by cleaning the cab data, removing dirty data, extracting the travel OD of passengers, and using singular value decomposition (SVD) to perform spatio-temporal dissection of the demand matrix, and combining with DBSCAN clustering to analyze the travel demand from and to the hubs. The spatio-temporal distribution characteristics are used to identify the potential demand for demand-responsive feeder transit. The results show that the method can accurately identify demand-responsive feeder transit demand, which is important for attracting a shift from private cars to public transportation.