An Artificial Neural Network Model for Highway Accident Prediction: A Case Study of Erzurum, Turkey
Abstract
This study presents an accident prediction model of Erzurum’s Highways in Turkey using artificial neural network (ANN) approaches. There are many ANN models for predicting the number of accidents on highways that were developed using 8 years with 7,780 complete accident reports of historical data (2005-2012). The best ANN model was chosen for this task and the model parameters included years, highway sections, section length (km), annual average daily traffic (AADT), the degree of horizontal curvature, the degree of vertical curvature, traffic accidents with heavy vehicles (percentage), and traffic accidents that occurred in summer (percentage). In the ANN model development, the sigmoid activation function was employed with Levenberg-Marquardt algorithm. The performance of the developed ANN model was evaluated by mean square error (MSE), the root mean square error (RMSE), and the coefficient of determination (R2). The model results indicate that the degree of vertical curvature is the most important parameter that affects the number of accidents on highways.
References
World Health Organization. Global Plan for the Decade of Action for Road Safety 2011-2020.[Internet] [cited 2012 Dec 5].Available from: http://www.who.int/roadsafety/decade_of_action/plan/plan_english.pdf.
Maher MJ, Summersgill J.A comprehensive methodology for the fitting of predictive accident models. Accident Analysis and Prevention. 1996;28(3):281-296.
Miaou SP. The relationship between truck accidents and geometric design of road sections: Poisson versus negative binomial regressions. Accident Analysis and Prevention. 1994;26(4):471-482.
Miaou SP, Lum H. Modelling vehicle accidents and highway geometric design relationships. Accident Analysis and Prevention. 1993;25(6):689-709.
Lee JY, Chung JH, Son B. Analysis of traffic accident size for Korean highway using structural equation models. Accident Analysis and Prevention. 2008;40:1955-1963.
Persaud B, Dzbik L. Accident prediction models for freeways. Transportation Research Record. 1993;1401:55-60.
Caliendo C, Guida M, Parisi A. A crash-prediction model for multilane roads. Accident Analysis and Prevention. 2007;39:657-670.
Knuiman MW, Council FM, Reinfurt DW. Association of median width and highway accident rates. Transportation Research Record. 1993;1401.
Fridstrøm L, Ifver J, Ingebrigtsen S, Kumala R, Krogsgard Thomsen L. Measuring the contribution of randomness, exposure, weather, and daylight to the variation in road accident counts. Accident Analysis and Prevention. 1995;27:1-20.
Hadi MA, Aruldhas J, Chow L, Wattleworth JA. Estimating safety effects of cross-section design for various highway types using Negative Binomial regression. Transportation Research Record. 1995;1500.
Persaud B, Retting RA, Lyon C. Guidelines for the identification of Hazardous Highway Curves. Transportation Research Record 2000;1717:14-18.
Abdel-Aty MA, Essam Radwan EA. Modelling traffic accident occurrence and involvement. Accident Analysis and Prevention. 2000;32:633-642.
Hauer E. Statistical road safety modelling. Proceedings of the 83rd TRB Annual Meeting; 2004 Jan 11-15; Washington, D.C., USA.
Hauer E. Safety models for urban four-lane undivided road segments. Proceedings of the 83rd TRB Annual Meeting; 2004 Jan 11-15; Washington, D.C., USA.
Chang L. Analysis of freeway accident frequencies: Negative binomial regression versus artificial neural. Safety Science. 43(8):541-557. doi:10.1016/j.ssci.2005.04.004
Kalyoncuoglu SF, Tigdemir M. An alternative approach for modelling and simulation of traffic data: artificial neural networks. Simul Model Pract Theory. 12(5):351-362. doi:10.1016/ j.simpat.2004.04.002.
Akgüngör AP, Doğan E. Estimating road accidents of Turkey based on regression analysis and artificial neural network approach. Advances in Transportation Studies International Journal. 2008 Nov;16:11-22.
Akgüngör AP, Doğan E. An artificial intelligent approach to traffic accident estimation: model development and application. Transport.2009; 24(2):135-142. doi:10.3846/1648-4142. 2009
Akgüngör AP, Doğan E. An application of modified Smeed, adapted Andreassen and artificial neural network accident models to three metropolitan cities of Turkey. Scientific Research and Essays. 2009 Oct;4(9):906-913.
Cansız OF. Improvements in estimating a fatal accidents model formed by an artificial neural network. Simulation. 2011;87(6):512-522. doi:10.1177/0037549710370842
Çodur MY. Traffic Accident Prediction Models: Applications for Surrounding Highways of Erzurum [PhD Thesis] [in Turkish]. Graduate School of Natural and Applied Sciences, Department of Civil Engineering, Ataturk University; 2012.
Narendra KS, Parthasarathy K. Identification and control of dynamical systems using neural networks. IEEE Transaction Neural Networks. 1990;1(1):4-27.
Narendra KS, Mukhopadhyay S. Adaptive control using neural networks and approximate models. IEEE Transactions on Neural Networks. 1997 May;8(3):475-485.
Singh P, Deo MC. Suitability of different neural networks in daily flow forecasting. Applied Soft Computing. 2007;7(3):968–978.
Ertuğrul S, Hizal NA. Neuro-fuzzy controller design via modelling human operator actions. Journal of Intelligent Fuzzy Systems. 2005;16:133-140.
Hamed M, Khalafallah M, Hassanien E. Prediction of Wastewater Plant Performance Using Artificial Neural Networks. Environmental Modelling and Software. 2004;19(10):919-928.
Svetoslavova Kovacheva Z. Application of Neural Networks to Data Mining. SQU Journal for Science. 2007;12(2):75-85.
Alyuda Neuro Intelligence Manual; 2004.Available from: http://www.alyuda.com
Bates JJ, Quarmby DA. An economic method for car ownership forecasting in discrete areas. Mathematical advisory unit note 219. London: Department of the Environment; 1971.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
