A Spatial-Temporal Kernel Density Estimate Method for Traffic Violation Hotspot Identification

Identifying traffic safety hotspots is an effective countermeasure to finding the most hazardous sites from massive road networks. In this paper, a spatial-temporal kernel density estimate model is presented to identify and analyze the traffic violation hotspots at urban intersections. Data from 200 days, which included 52 intersections and 110,173 violations in Wujiang, China, were utilized to carry out the empirical analyses. Three main types of violation data were mapped by using three-dimensional cubes, which clearly showed traffic violation hotspots with the vertical color of every pillar cluster. Compared with the traditional two-dimensional hotspot maps, the STKDE vividly showed space and temporal clusters of violation hotspots. According to the results, hotspots identified by times of day are found to be mostly different from each violation type. These findings provide insights into the temporal effects of risk factors and can support the development of spatial-temporal dependent safety countermeasures.