Many-to-one stable matching for taxi-sharing service with selfish players

Abstract

In taxi-sharing, matching problems can be divided into two parts, i.e., passenger-to-taxi matches and passenger-to-passenger matches. Considering of the selfishness of passengers and taxi drivers, attention needs to be devoted to their preferences over matching partners. In this paper, the concept of the scheme is introduced as a medium to formulate the preferences of passengers and taxi drivers. A stable matching framework is first designed to model the matching problem of taxi-sharing service with peer effect. A branch-and-price algorithm is tailored to solve the many-to-one stable matching problem. Dalian taxi data are used to assess the performance of the proposed model and algorithm. The results indicate that the variable discount strategy is a useful tool to balance the demand and supply. The matching rate of taxis increases by an average of 5.39% compared with the fixed discount strategy. Furthermore, when the taxi-sharing reaches a high level of participation, the benefit received from expanding the service space of taxis from a 3-min drive to a 10-min drive is limited. However, a relaxed preference rule could increase the matching rate of passengers by 25.6%. When the passengers’ preferences over co-riders are not considered as the traditional preference rule, the waiting time of passengers increases by 77.1%.