Time-evolving O-D matrix estimation using high-speed GPS data streams
作者：
Highlights：
• We proposed a data driven method to incrementally build O-D matrices.
• The Regions of Interest on this matrix are defined using spatial clustering.
• The sample’s attributes are discretized using a novel multidimensional hierarchy.
• The target’s p.d.f. is approximated using multidimensional histograms.
• This methodology is validated using 1 million of real-world trips.
摘要
•We proposed a data driven method to incrementally build O-D matrices.•The Regions of Interest on this matrix are defined using spatial clustering.•The sample’s attributes are discretized using a novel multidimensional hierarchy.•The target’s p.d.f. is approximated using multidimensional histograms.•This methodology is validated using 1 million of real-world trips.
论文关键词：Online machine learning,Data streams,Mass-based clustering,Incremental discretization,Urban mobility analysis,Origin-destination matrix
论文评审过程：Received 4 November 2014, Revised 27 August 2015, Accepted 29 August 2015, Available online 11 September 2015, Version of Record 10 November 2015.
论文官网地址：https://doi.org/10.1016/j.eswa.2015.08.048
