Ridesharing with social contacts (i.e., ‘friends’) is substantially more accepted than with strangers. However, limiting ridesharing to friends while rejecting strangers also reduces ride choices and increases detour costs. This work studies, from a theoretical perspective, whether the additional detour costs of limiting shared rides to social network contacts would be prohibitive. It proposes a social network based ridesharing algorithm with heterogeneous detour tolerances for varied social contacts. The theoretical matching rates and detour costs are compared in a simulation for three levels of social connectivity: travelling with direct contacts only, with direct and indirect contacts, or with anyone. The simulation allows for a systematic and comprehensive testing of system behaviour when varying the parameters of social network structure, detour tolerance, and spatial distribution of friendship. Results show that for a clustered friendship – the expected spatial distribution of a social network growing with a ride-sharing network – ridesharing with friends does not cause significantly higher costs. Furthermore, the algorithm prioritising friends can substantially increase the matching of friends. An empirical study justifies the findings.
The growing amount of cars on the road results in increasing congestion of urban traffic, which then leads to higher fuel consumption, longer average travel time, more environmental pollution, and less patience of people with their daily travel (International Transport Forum, 2013). One known way to reduce the traffic load on the road without harming accessibility is looking for higher occupancy rates per vehicle, i.e., through ridesharing. Trajectory analysis has indicated good chances of ridesharing according to space-time concurrence (Santi et al., 2014). But despite the environmental and economic benefits, there is still a low rate in participation of ridesharing (Amey, 2010; Chaube, Kavanaugh, & Pérez-Quiñones, 2010; Wessels, 2009). The contradiction is partially due to the strong reluctance to share rides with strangers according to some surveys (Chaube et al., 2010; Wessels, 2009). The low willingness for ridesharing with strangers signifies that many of the existing ride opportunities, according to trajectory overlap, are actually inacceptable for a certain person. The preference for travelling with social contacts (a first or second degree socially connected person, hereafter called “friend”) seems to be a good reason to limit ridesharing to friends. Realising this argument, Facebook filed recently a patent called “Event-based ride-sharing” (Richardson, Petrescu, & Finch, 2016), which allows drivers to select users based on his/her social network connections for an online negotiation on carpooling to an event.
However, a ridesharing system limited to friends has yet to address the impact of missing chances of getting a ride. Ridesharing exclusively with friends while declining offers from strangers may lead to fewer opportunities to get a ride within a given space-time budget and to higher detour costs. This study aims to examine the theoretical costs and benefits of ridesharing with friends by systematic and comprehensive variation of parameters in a simulation beyond a particular context. Empirical tests are run as validations. As social networks are gaining attention in travel behaviour research (Arentze & Timmermans, 2008; Hackney & Marchal, 2011), this work helps gain insightful understanding into how spatial structures of social networks affect ridesharing results.
Please, read the full article at: http://www.sciencedirect.com/science/article/pii/S0198971516303696