Prediction of silent users of car-sharing based on Logistic Regression Model

Abstract

Car-sharing is a new transportation mode with the rapid development of mobile Internet. It is necessary for the operation companies to keep the number of carsharing users and intelligent transportation can be realized by discovering the users before they loss and preventing the loss. In this paper, based on the massive order data of a car-sharing company in Beijing, the behavior of the users is analyzed before they become silent. In the light of Logistic Regression Model, the loss rate of users in next month is predicted based on the car-renting behavior in the former three months, and some suggestions are proposed to prevent the loss. The result shows that there is a significant difference in the behavior between the lost users and non-lost users. The accuracy of the prediction is 97.9%. The method proposed in the paper provides a theoretical basis for enterprises to deal with the early warning and recall of the lost users.