Abstract

Discovering Spatial Structure in Massive Mobility Data
Track: Education and Training
Authors: Diansheng Guo, Xi Zhu, Peng Gao, Hai Jin

Human mobility and spatial interaction data have become increasingly available due to the wide adoption of location-aware technologies such as GPS devices and cell phones. Examples of mobility data include human daily activities, vehicle trajectories, taxi trips, animal movements, among many others. In this paper we focus on the discovery of the underlying spatial structure from mobility data such as functional regions (where there are more connections within regions than between regions). Specifically, our approach involves two steps: spatial clustering and contiguity constrained graph partitioning. The spatial clustering step is to aggregate locations (e.g., millions of GPS points) into small spatial clusters (areas) of similar size. The second step constructs a directed graph with connections (e.g., trips) among the spatial clusters and then partitions the graph to discover a hierarchy of functional regions. We present a case study with a large dataset of taxi trajectories to demonstrate the methodology.