|Title||The Tradeoff Between Efficiency and Fellow Passenger Preference: a Preference-Based Ridesharing Model|
|Publication Type||Conference Paper|
|Year of Publication||2017|
|Authors||Hongmou Zhang, Jinhua Zhao|
|Conference Name||Transportation Research Board 96th Annual Conference|
|Conference Location||Washington, D.C.|
|Keywords||Manhattan, matching, preference, ridesharing|
Advances in information technology are enabling the growth of real-time ridesharing—whereby passengers are paired up on car trips to improve system efficiency by using fewer cars. Lesser known, however, are the opportunities of shared mobility as a tool to foster and strengthen human interactions. The nature of shared car rides is impromptu, captive for a considerable duration, and remarkably more intimate, representing a unique juxtaposition of spontaneity and intensity. While ridesharing services optimize the matching of fellow passengers based on efficiency criteria, like maximal paired trips, or minimal VMT, they do not consider passengers’ preference for each other—or only use it as a restriction—especially for the preference from sociodemographic characteristics, like gender. We propose a preference-based matching model, which optimizes fellow passenger pairing by using the most stably favorable fellow passengers as a system objective, and evaluate the tradeoff between it and efficiency-based matching models. In the model, we implement two types of synthetic preference: 1) random preference, to show efficacy of the model and the range of tradeoff between different matching schemes, and 2) group-based preference of five scenarios, to illustrate how preference coming from sociodemographic characteristics lead to different matching outcomes. We use the taxi trips of Manhattan to put this model in real-world scenarios. The results of the paper show that compared to the matchings under different efficiency-based optimization goals, the preference-based model can increase the average ranking of paired fellow passenger to the top of preference lists at the compensation of only a moderate amount of efficiency loss. The model can be used for ridesharing service design, and as a reference for policy makers of ridesharing regulation.