Recommending Profitable Taxi Travel Routes based on Big Taxi Trajectory Data
Abstract
Recommending potential profitable routes to reduce the cruising distance of taxis is an active research topic.
This thesis first introduces a temporal probability grid network generated from taxi trajectories, where each grid has been assigned two temporal properties: probability and capacity. Two profitable route recommendation algorithms are proposed: the Shortest Expected Cruising Route (SECR) and Adaptive Shortest Expected Cruising Route (ASECR) algorithms. The ASECR algorithm is an extension of SECR that updates profitable routes constantly, and renews the temporal probability grid network dynamically. To handle large amounts of trajectory data and improve the efficiency of constantly updating the routes, a new data structure kdS-tree with a MapReduce model is proposed.
Case studies compare the time and distance performances of the ASECR and SECR algorithms with two other methods, i.e. the LCP method and the baseline method. These comparisons demonstrate the effectiveness and efficiency of the two proposed algorithms.
Description
Keywords
Computer Science, Geotechnology
Citation
Yang, W. (2015). Recommending Profitable Taxi Travel Routes based on Big Taxi Trajectory Data (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/27559