Multi-source data-driven prediction for the dynamic pickup demand of one-way carsharing systems

Abstract

The one-way carsharing system has been widely used in the carsharing field due to its flexibility. However, one of its main disadvantages is the imbalance between supply and pickup demand. At present, multi-source data are available for the real-time prediction of pickup demand. The multi-source data that are used for this purpose include real-time user application log data, historical order data, real-time station data, and user characteristic data. Based on these data, a demand prediction model was used to predict in real-time whether there is a pickup demand, and a demand time prediction model was applied to forecast the time at which a sharing vehicle is needed. Finally, a case study was conducted using 10 stations’ one-week field data to test the benefits of the models. The potential application of this study would effectively guide the system to formulate an active operation optimisation strategy to meet users’ demand.