Predicting carsharing station-based trip generation using a growth model

Carsharing is a service that allows members to rent cars for a limited time. In Montreal, Quebec, Canada, two types of services exist: a station-based and a free-floating service. This paper proposes a trip generation model for the station-based service of the Communauto carsharing operator for 2016. To better understand relations between space and time, a growth model is used, considering these factors at different levels. For example, some factors can impact all stations similarly, while other factors may impact each station differently. Thus, this model allows to consider both spatial and temporal variables allowing more precise estimations. The aim of this research is to estimate carsharing trip generation at the station level and provide insights into the impacts of implementing new stations on demand. A step-by-step approach was adopted to define the best predictive model for the use of carsharing stations. While more complex model formulations need to be tested to enhance the analysis, the final growth model obtained indicates that, in addition to the number of vehicles available at the stations, several exogenous factors have a significant impact on the trip generation rate of a carsharing station. For instance, the model shows that demographic factors, walkability level and number of bus stations have significant impacts on the use of carsharing stations.

Uncontrolled Keywords

carsharing; station-based; trip generation; growth model; multilevel analysis; multivariate approach