A Case Study on Kernel Density Estimation and Hotspot Analysis Methods in Traffic Safety Management

Identification of hotspots is the foremost exercise in accident prevention which enhances the efficiency of traffic safety management system. The aim of the paper is to identify and rank the accident hotspots based on Kernel Density Estimation (KDE) and Hotspot Analysis (Getis-Ord Gi
^*
) methods. KDE provides the visual representation of the hotspots and the Hotspot Analysis (Getis-Ord Gi
^*
) provides the statistical significance of the hotspots in terms of Getis-Ord (G
_i
^*
) statistic. In this study, Des Moines City in Polk county of the United States is considered as the study area for application of KDE and G
_i
^*
methods, for illustration purposes. Five years of crash data from 2008 to 2012 are used in the study. Firstly, a visual representation of the hotspots is obtained from KDE. Then, Hotspot Analysis (Getis-Ord Gi*) is performed to test the statistical significance for hotspot locations identified from KDE. The statistically significant hotspots are then ranked based on the density estimate. The methodology could be applied for any study area including developing countries with sufficient accident data for hotspot identification and prioritization of actions to reduce traffic accidents.