Balancing efficiency and fairness in ridesharing dispatching

Abstract

Ridesharing Apps have increased its significance in urban mobility that provides convenience and higher flexibility to match riders and on-demand drivers with immediate response time. Traditional systems for taxis dispatching were designed to match the driver to a request in terms of shortest detour distance or minimal waiting time, without considering other factors, which may affect the overall matching efficiency. In recent years, different scheduling algorithms were proposed to address trade-offs among waiting time, detour, profits, and costs, etc. to tailor the dynamic needs of passengers. However, there is yet a solution that can balance on drivers’ benefits and fairness. In this project, drivers’ fairness, which specifically refers to the discrepancy of their daily profits, was taken into consideration. Corresponding efficient matching algorithms were designed to balance the trade-off between dispatching efficiency and drivers’ fairness. Instead of assigning the driver simply based on shortest detour or shortest waiting time, the algorithm takes the historic profits earned by on-demand drivers as a partial weight for matching, and evaluate an innovative index - Balanced Satisfaction Index (BSI). Historic large-scale Chicago Taxi Trip dataset containing average 30,000 daily trip records were used for testing of algorithm performance. Results indicate a significant improvement of driver’s fairness with a range of 60% to 85% given order range between 2,000 to 30,000. While Efficiency is only slightly impaired. Other analyses on Radius and weight parameter were also conducted. That being said, taking into consideration of drivers’ fairness does not impose a serious effect on overall efficiency, while playing an important role in guaranteeing that each driver can receive similar amounts of order every day. As the principle of fairness is vital for an inclusive workplace, this project provides a useful reference for ridesharing companies to promote the fairness among their designated drivers, to realize a less cut-throat competitive but more harmonious ridesharing environment.