Recommending Pick-up Points for Taxi-drivers Based on Spatio-temporal Clustering
Authors:
- Mingyue ZhangMZMingyue Zhang
- Jianxun LiuJLJianxun Liu
- Yizhi LiuYLYizhi Liu
- Zhenyang HuZHZhenyang Hu
- Liang YiLYLiang Yi
- 1 November 2012
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 67-72
Abstract
Using GPS trajectories to recommend pick-up points for taxi driver comes to be a promising approach of increasing profits and decreasing pollutions. In the existing methods, nearly all the GPS data of a city are computed for recommendation. However, it is time-consuming and not accurate enough owing to too much spatio-temporal noise. Therefore, we propose a novel method of recommending pick-up for taxi driver based on spatio-temporal clustering. It is made up of data preprocessing and real-time recommendation. Firstly, we capture the historical pick-up points by analyzing their intervals. These points are clustered at different time and different regions to create candidate pick-up points. Secondly, after ranking the candidate pick-up points around the taxi, the top-5 valuable pick-up points are recommended for taxi drivers. The experimental results, whose data come from Microsoft Research Asia, show that our method can effectively recommend pick-up points for taxi drivers.Keywords
This publication has 11 references indexed in Scilit:
- A Taxi Driving Fraud Detection System2011
- Where to find my next passenger2011
- A taxi business intelligence system2011
- Driving with knowledge from the physical world2011
- T-drive2010
- Mining interesting locations and travel sequences from GPS trajectories2009
- Understanding mobility based on GPS data2008
- Bayesian hierarchical clustering2005
- Using GPS to learn significant locations and predict movement across multiple usersPersonal and Ubiquitous Computing, 2003
- An efficient k-means clustering algorithm: analysis and implementationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2002