Route recommendation method for frequent passengers in subway based on passenger preference ranking

Abstract

To provide personalized and precise guidance services to subway passengers, a route recommendation method for frequent subway riders based on passenger preference ranking is proposed. Firstly, a data-driven analysis method of travel preference is proposed to identify the travel preference of frequent passengers in different origin-destinations (ODs) and time periods. Secondly, a route recommendation model of dictionary order preference based on just noticeable difference (JND) is proposed to solve the problem of different passenger perception in route attributes. Thirdly, the model of depth deterministic strategy gradient (DDPG) is proposed to consider dynamic parameters of JND. A case study of Guangzhou Metro is illustrated to verify the accuracy of the route recommendation model. The results show that the individual passenger recommendation in the proposed method is more accurate than the traditional logit model and the JND model without parameter optimization by 30% and 20%. The group recommendation in the proposed method is more accurate than the comparative models by 37% and 11%, respectively. Specifically, the accuracy of this method in group recommendation fluctuates less, and its accuracy variance is much smaller than the traditional logit model. Therefore, the method proposed in this study performs better in reflecting personalized need of passenger travel.