Context-aware big data analytics and visualization for city-wide traffic accidents

Abstract

Various traffic big data has been emerging in cities, such as road networks, GPS trajectories of buses and taxicabs, traffic flows, accidents, etc. Based on the massive traffic accident data from January to December 2015 in Xiamen, China, we propose a novel accident 0analytics and visualization method in both spatial and temporal dimensions to predict when and where an accident with a specific crash type will occur consequentially by whom. First, we analyze and visualize accident occurrences and key features in both temporal and spatial view. Second, we propose our context-aware methodology. Finally, we illustrate spatio-temporal visualization results through two case studies. These findings would not only help traffic police department implement instant personnel assignments among simultaneous accidents, but also inform individual drivers about accident-prone sections and most dangerous time spans, which would require their most attention.