An Entropy-based Model for Recommendation of Taxis' Cruising Route

EasyChair Preprint 1461

Abstract

Recommending an optimal cruising route for a taxi-driver helps he/she save the taxi’ idle running time, which can then improve the taxi-drivers’ income or reduce the taxi’s energy consumption. Mining the optimal knowledge for recommendation from the vast previous drivers’ GPS trajectories is a possible way since the trajectories are now easily recorded and kept in databases. Lots of work have been done then. However, existing methods mostly recommend pick-up points for taxis only. Their performance is not good enough since there lacks a good evaluation model for the pick-up points selected. In this paper, we propose a novel evaluation model based on information entropy theory for taxis’ cruising route recommendation. Firstly, we select more positional attributes from historical pick-up points in order to obtain accurate spatial-temporal features. Secondly, an integrated evaluation model learning from historical pick-up points is constructed based on the information entropy theory, which is applied to get the future pick-up points. We then design a pruning algorithm to recommend a series of successive points to generate a cruising route for a taxi driver. Experiments are done on a real dataset and the results show that our method can significantly improve the recommendation accuracy of pick-up points, and help taxi-drivers make profitable benefits more than before.

Keyphrases: Location Based Service (LBS), Taxis' cruising route recommendation, Trajectory Data Mining, information entropy, services recommendation

@booklet{EasyChair:1461,
  author    = {Yizhi Liu and Xuesong Wang and Jianxun Liu and Jianjun Wang and Zhuhua Liao and Yijiang Zhao},
  title     = {An Entropy-based Model for Recommendation of Taxis' Cruising Route},
  howpublished = {EasyChair Preprint 1461},
  year      = {EasyChair, 2019}}