This study aims to improve the accuracy of a convolutional neural network (CNN) based model. That predicts the likelihood of accidents on a specific road section from the present to 2 h in the future using a wide range of temporal and spatial sensor information developed in previous studies as input to reduce accidents. In addition to previous studies that only used traffic data (i.e., speed, traffic volume, time occupancy, etc.), we considered time data (i.e., day of the week, time of day, etc.) and weather data as additional explanatory variables. Then, using the chi-square test, we selected the information that contributed to improving the accuracy of accident occurrence prediction and added it as input to the CNN-based model. Compared with the base model, the average F1-score of the proposed model was improved by 19.1%.
Improving the Accuracy of Traffic Accident Prediction Models on Expressways by Considering Additional Information
Int. J. ITS Res.
International Journal of Intelligent Transportation Systems Research ; 20 , 1 ; 309-319
2022-04-01
11 pages
Article (Journal)
Electronic Resource
English
Traffic Management for Urban Expressways
British Library Conference Proceedings | 1994
|Multisource traffic information fusion method for expressways
European Patent Office | 2015
|Accident rates on German expressways in relation to traffic volumes and geometric design
Engineering Index Backfile | 1957
|European Patent Office | 2023
|