Prediction model of demand for public bicycle rental based on land use

Abstract

Land use is a primary factor affecting the demand for public bicycle rentals. Demand for public bicycle rentals during different periods of time were predicted using the following procedures. First, walking distances from the rental stations where riders returned the public bicycles to the final destinations were obtained by field investigation, and the 85th percentile statistical values were used as the scopes of influence of those stations. Then, a relationship model among the rental demands for public bicycles and the features of land use inside the influence scope of the rental station was established based on a linear regression model. Finally, considering the public bicycle system in the old urban region of Zhenhai in Ningbo city, the newly established prediction model for rental demand was tested. Results show that the model can predict the daily rental demand, rental demand during the morning peak, returns during the morning peak, rental demands during the evening peak, and returns during the evening peak. The demand prediction model can provide a significant theoretical basis for preparing the layout stations, operation and management strategies, and vehicle scheduling in the public bicycle system.