Abstract:

In order to improve the carpooling matching probability and satisfaction, this paper proposes a dynamic carpooling matching method that considering passengers’ time-price elasticity. RP+SP questionnaire survey method was applied to collect passengers’ individual attributes and carpooling choosing behavior under different travel scenarios. Clustering passengers based on the carpooling preferences, passengers were divided into three categories. A discrete elasticity analysis model was established to obtain the time-price elasticity of carpooling and non-carpooling  passengers for the three categories. Incorporating time-price elasticity into travel costs, a generalized cost function was established for all passengers and drivers. A two-level planning model was constructed, for which, the upper-level model considered the benefits of drivers and passengers, while the lower-level model aimed to maximize the carpooling probability of all passengers. Considering route and capacity constraints, a carpooling matching algorithm was designed. Finally, taxi trajectory data of Xi'an was took as an example, different carpooling matching schemes were solved when considering different time-price elasticity, and the differences between all matching schemes were compared. The results indicate that when passengers have high time-price elasticity (2.22, 0.99), the non-carpooling costs are lower than carpooling costs, and passengers cannot effectively match to each other. When passengers have low time-price elasticity (0.12), multiple passengers can be matched to implement carpooling. The conclusion of this study demonstrates the necessity of time-price elasticity and the results can be used to guide government for carpooling order allocation and taxi dispatching.

Key words: urban transportation, carpooling matching, elasticity analysis, two-level planning model