This study aims to develop classification models capable of identifying real-time traffic conditions prior to sideswipe crashes. A machine learning algorithm named support vector machine (SVM) was applied as the classifier. Historical loop detector data for sideswipe crashes and corresponding non-crash cases were collected from Interstate 894 in the Milwaukee, Wisconsin, United States. The collected detector data were aggregated into three sets of traffic parameters, all of which describe variances between adjacent lanes. Each set of traffic parameters were then explored as potential inputs into SVM classifiers respectively. Classification results showed that averages of volume, speed, and occupancy combined with absolute differences of volume, speed, and occupancy between adjacent lanes provide the best classification accuracy, specifically, an 84.6% identification of sideswipe crashes and non-crash cases. The finding in this study also indicated that SVM is an effective tool in identifying sideswipe crash prone traffic conditions.
Identification of Traffic Conditions Leading to Sideswipe Crashes on Freeways
11th International Conference of Chinese Transportation Professionals (ICCTP) ; 2011 ; Nanjing, China
ICCTP 2011 ; 2092-2101
2011-07-26
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Potential Real-Time Indicators of Sideswipe Crashes on Freeways
Transportation Research Record | 2006
|Potential Real-Time Indicators of Sideswipe Crashes on Freeways
Transportation Research Record | 2006
|Potential Real-Time Indicators of Sideswipe Crashes on Freeways
British Library Conference Proceedings | 2006
|Potential Real-Time Indicators of Sideswipe Crashes on Freeways
Online Contents | 2006
|Characteristics of Traffic Crashes on Freeways in China
Transportation Research Record | 2007
|