Optimizing Locations for a Vehicle Sharing System

Car-sharing business provides members access to a fleet of shared-use vehicles in a network of locations on a short-term, as-needed basis. It allows individuals to gain the benefits of private vehicle use without the costs and responsibilities of ownership. The primary advantage of car sharing as opposed to the car rental business is that it provides flexibility of using the vehicles for shorter periods of time. Secondly if you are a registered customer of the company, the process of accessing and using the vehicle is quite simple, self-serving and hassle-free. One of the main problems faced by the car-sharing businesses is that of finding the best stations to place the facility. These locations should be chosen by attractiveness of various parameters such as the socio-demographic-economic profile of the population that resides or works at the location and accessibility of other forms of public transport in that location. However the presence of one or more stations in the vicinity and the type of vehicles placed in a station are also critical for decision-making. It must be noted that the attractiveness factor of a parameter would be shared between different stations if they are present in the sphere of influence of that parameter. This study analyzes the performance of an electric car sharing service across different stations in and around the city of Nice. This service is operated by a major public transport operator in France and commenced its operations in April 2011. The main measure of performance of the stations is the average number of rides (usages) per day. The objective of this study is two-fold one, to analyze the performance of the car-sharing service across all stations and estimate the key drivers of demand, and secondly, to use these drivers to identify future station locations, such that the overall system performance is maximized. The methodology used by us to optimize the station locations follows two steps. In the first step, we perform an extensive data analysis and determine all those factors that we expect to be driving the demand for the service and build a linear regression model with station performance as the dependent variable that is explained by a host of other independent variables, such as public transport ridership, share of car-users in the locality, share of high income / education groups in the locality population, population density and presence of other mobility generators such as hotels, colleges and commercial centers. In the second and final step, we use the attractiveness of the different localities to optimally locate the new stations for the service. The main trade-off decision made by the model involves locating more stations in highly attractive localities versus locating new stations in less attractive but untapped localities. We validated and reported our results on a set of new stations that were opened quite recently, after the commencement of our study.