Authors: Panini Chowdhury*, University of Toledo, Bayes Ahmed, University College London, Bhuiyan M. Alam, University of Toledo
Topics: Transportation Geography, Urban Geography, Planning Geography
Keywords: Traffic accidents,Developing Countries,multinomial logistic regression,Network Based Density Estimation,Hotspot analysis
Session Type: Poster
Start / End Time: 5:20 PM / 7:00 PM
Room: Napoleon Foyer/Common St. Corridor, Sheraton, 3rd Floor
Presentation File: Download
This paper demonstrates traffic accident situation in Dhaka Metropolitan Area, Bangladesh, covering 2716 accidents from 2007-2011. Hotspot and Inverse Distance Weighted methods helped identify the most vulnerable accident locations. Global Moran’s I was applied to understand the spatial autocorrelation of the accident distribution pattern. Network Kernel Density Estimation was applied to find the relationship of crash density with land use, while Multinomial Logistic Regression was applied to understand the impacts of built environment factors on accident severity. No cold spot in the study area was found indicating a uniform but high crash severity. Spearman’s correlation coefficient was applied to find the relationship between population density and accident hotspots. Commercial mixed zones are the worst sufferer of these accidents. The most accident-prone periods of the year institute March – June. Presence of road divider, daylight, and feeder road show positive impacts on reducing fatal accidents. Police controlled intersection points can significantly reduce major and minor injuries.