Dynamic Autoregressive Neural Networks for Spatially Distributed Time Series Prediction of Car Crashes in Urban Networks
Following the public and administrative concern on road safety that is evident in the last decades a significant effort has been done for surveying, modelling and predicting road accidents. One of the important aspects for optimal planning and scheduling of road safety resources stands for the prediction models. The majority of the existing models are aiming either on correlating traffic variables with crash occurrence or modelling crash frequencies both in highways as well in urban network links and intersections. In the current paper, results from an exploratory analysis are presenting, based on spatially distributed time series prediction modelling, belonging to the Artificial Intelligent class of modelling approaches. Such approach in road safety is not widely presented in the literature although can be regarded as useful in cases where detailed and reliable information of traffic characteristics is not available. In particular, the performance of Artificial Intelligent-based predictors is investigated. A suitable for such analysis realistic database (composed of daily records covering almost 6 years) from Riyadh, capital of the Kingdom of Saudi Arabia is analyzed by a hybrid dynamic Autoregressive Artificial Neural Network setup, providing evidence on the performance of such approaches in predicting spatially distributed car crashes time series.
-
Supplemental Notes:
- This paper was sponsored by TRB committee ABJ70 Artificial Intelligence and Advanced Computing Applications.
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Dimitriou, Loukas
- Hassan, Hany M
-
Conference:
- Transportation Research Board 92nd Annual Meeting
- Location: Washington DC, United States
- Date: 2013-1-13 to 2013-1-17
- Date: 2013
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; References;
- Pagination: 17p
- Monograph Title: TRB 92nd Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Databases; Intelligent transportation systems; Neural networks; Regression analysis; Time series analysis; Traffic crashes; Traffic flow; Traffic safety
- Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management; Safety and Human Factors; I70: Traffic and Transport; I81: Accident Statistics;
Filing Info
- Accession Number: 01476781
- Record Type: Publication
- Report/Paper Numbers: 13-4272
- Files: TRIS, TRB, ATRI
- Created Date: Mar 27 2013 9:37AM