A Microscopic Spatial-Temporal Forecast Framework for Inflow and Outflow Gap of Free-Floating Bike Sharing System

摘要

Free-floating bike-sharing system (FFBSS) has been popularized rapidly as a newly arisen short-distance transportation mode. However, over-delivery of bikes also brings many problems. Evaluating the usage of bikes is the basis for delivery and relocating. Since users play a significant role in the movement of the bike, it is necessary to consider the unlocking and locking events as the inflow and outflow of the area when evaluating the usage. Based on the consideration of both kinds of user behavior, the inflow and outflow gap of FFBSS is gridded by using different spatial-temporal parameters. The performances of linear regression (LR), support vector regression (SVR), random forest (RF), and gradient boost machine (GBM) are compared. The results show that GBM can get the best results in most of the time.