Capturing the characteristics of carsharing users based on a data-driven method

Abstract

In order to capture the characteristics of carsharing users, we applied a data-driven method to classify the users into typical clusters according to the contributions to the carsharing companies. The data used in this study are collected in the order information from the carsharing company in Beijing, China. The RFM model is employed to determine the influencing factors for clustering analysis. Accordingly, the most recent consumption, consumption frequency and spending are adopted to represent the user value. The users can thus be classified into three groups, named as potential users, high value users and the loss of users. It is expected that the results can improve the user management of the carsharing company.