MARIO: A spatio-temporal data mining framework on Google Cloud to explore mobility dynamics from taxi trajectories. (15th August 2020)

• Record Type: Journal Article
• Authors: Ghosh, Shreya; Ghosh, Soumya K.; Buyya, Rajkumar
• Is Part Of: Journal of network and computer applications. Volume 164(2020)
• Publication Date: 2020-08-15
• Subjects: Trajectory -- Spatio-temporal data analysis -- Association rule mining -- Deep learning -- Travel demand prediction; Microcomputers -- Periodicals; Computer networks -- Periodicals; Application software -- Periodicals; Micro-ordinateurs -- Périodiques; Réseaux d'ordinateurs -- Périodiques; Logiciels d'application -- Périodiques; Application software; Computer networks; Microcomputers; Periodicals; 004.05; 004
• DOI: 10.1016/j.jnca.2020.102692
• Languages: English
• ISSNs: 1084-8045

Abstract

Abstract: With the major advances in location acquisition techniques, deployment of GPS enabled devices and increasing number of mobile users, substantial amount of location traces are generated from different geographical regions. It provides unprecedented opportunities to analyze and derive valuable insights of urban dynamics, specifically, time-dependent mobility patterns and region-specific travel demands . This work proposes an end-to-end mobility association rule mining framework called MARIO, conducive to extract urban mobility dynamics through analysing large taxi trip traces of a city. The MARIO framework consists of (i) generating mobility-dynamics network by spatio-temporal analysis of taxi-trips, (ii) finding travel demand variations in different functional regions of the urban area, (iii) extracting mobility association rules and (iv) predicting travel demands and traffic dynamics using extracted associative rules. The proposed MARIO framework is implemented in Google Cloud Platform and an extensive set of experiments using real GPS trace dataset of NYC Green and Yellow Taxi trace, Roma Taxi Dataset and San Francisco Taxi Dataset have been carried out to demonstrate the effectiveness of the framework. The performance of the proposed approach is significantly better than the baseline methods in predicting travel demands (with the reduction of average MAPE value and execution time by 50%). Highlights: Developing mobility association rule miner framework named MARIO over Google Cloud Platform. Modelling city dynamics network considering crowd behaviour, traffic flow and travel-demand of different functional regions. Proposing a novel hash-based indexing scheme to store and effectively retrieve trajectory data. Analysing the impact or effect of mobility events of one region to other regions using deep LSTM architecture. Predicting travel demand in different regions of a city efficiently and timely manner.