study of the severity of accidents in tehran using statistical modeling and data mining techniques

backgrounds and aims: the tehran province was subject to the second highest incidence of fatalities due to traffic accidents in 1390. most studies in this field examine rural traffic accidents, but this study is based on the use of logit models and artificial neural networks to evaluate the factors that affect the severity of accidents within the city of tehran. materials and methods: among the various types of crashes, head-on collisions are specified as the most serious type, which is investigated in this study with the use of tehran’s accident data. in the modeling process, the severity of the accident is the dependent variable and defined as a binary covariate, which are non-injury accidents and injury accidents. the independent variables are parameters such as the characteristics of the driver, time of the accident, traffic and environmental characteristics. in addition to the prediction accuracy comparison of the two models, the elasticity of the logit model is compared with a sensitivity analysis of the neural network. results: the results show that the proposed model provides a good estimate of an accident's severity. the explanatory variables that have been determined to be significant in the final models are the driver’s gender, age and education, along with negligence of the traffic rules, inappropriate acceleration, deviation to the left, type of vehicle, pavement conditions, time of the crash and street width. conclusion: an artificial neural network model can be useful as a statistical model in the analysis of factors that affect the severity of accidents. according to the results, human errors and illiteracy of drivers increase the severity of crashes, and therefore, educating drivers is the main strategy that will reduce accident severity in iran. special attention should be given to a driver’s age group, with particular care taken when they are very young. references ayati e. the cost of traffic accidents in iran. mashhad: ferdowsi university of mashhad; 2008. tehran safe community. 2012; available from: www.tehransafe14.com/default.aspx? 04&relatedid;=vnehvndy. pei x, wong s c, sze n.n. the roles of exposure and speed in road safety analysis. accid anal prev. 2012; 48: 464-71. national highway traffic safety administration, traffic safety facts fars/ges annual report, proceedings of 90th 2011; washington, d.c. jones a p, jorgensen s h. the use of multilevel models for the prediction of road accident outcomes. accid anal prev. 2003; 35(1): 59-69. miltner e, salwender h j. influencing factors on the injury severity of restrained front seat occupants in car-to-car head-on collisions. accid anal prev. 1995; 27(2): 143-50. srinivasan k k. injury severity analysis with variable and correlated thresholds: ordered mixed logit formulation. transp res rec. 2002; 1784: 132-42. garder p. segment characteristics and severity of head-on crashes on two-lane rural highways in maine. accid anal prev. 2006; 38(4): 652-61. zuxuan d, ivan j.n, gårder p. analysis of factors affecting the severity of head-on crashes two-lane rural highways in connecticut 2006; 41: 137-46. mercier cr, shelley ii mc, rimkus j b, mercier j m. age and gender as predictors of injury severity in head-on highway vehicular collisions.1997; 24: 37-46. haleem k, abdel-aty m. examining traffic crash injury severity at unsignalized intersections. j safety res. 2010; 41(4): 347-57. peek-asa c, britton c, young t, pawlovich m, falb s. teenage driver crash incidence and factors influencing crash injury by rurality. j safety res. 2010; 41(6): 487-92. kononen d w, flannagan c a c, wang s c. identification and validation of a logistic regression model for predicting serious injuries associated with motor vehicle crashes. accid anal prev. 2011; 43(1): 112-22. rifaat s m, tay r, de barros a. effect of street pattern on the severity of crashes involving vulnerable road users. accid anal prev. 2011; 43(1): 276-83. chong m, abraham a, paprzycki m. traffic accident analysis using machine learning paradigms. inform 2005; 29(1): 89-98. delen d, sharda r, bessonov m. identifying significant predictors of injury severity in traffic accidents using a series of artificial neural networks. accid anal prev. 2006; 38(3): 434-44. abdelwahab ht, abdel-aty ma. development of artificial neural network models to predict driver injury severity in traffic accidents at signalized intersections2001. hensher da, rose jm, greene wh. applied choice analysis: a primer: cambridge university press 2005. train ke. discrete choice methods with simulation. united states of america cambridge university press 2007. pal sk, mitra s. multilayer perceptron, fuzzy sets, and classification, ieee transactions on neural networks 1992. baykan n a, yilmaz n a. mineral classification system with multiple artificial neural networks using k-fold cross validation. j comput math. 2011; 16(1): 22-30. izenman aj. modern multivariate statistical techniques, springer science and business media 2008. elvik r. speed and road safety: synthesis of evidence from evaluation studies. transp res rec. 2005; 1908: 59-69. shams m, rahimi-movaghar v. risky driving behaviors in tehran, iran. traffic inj prev.2009; 10(1): 91-4. abdelwahab ht, abdel-aty m a. development of artificial neural network models to predict driver injury severity in traffic accidents at signalized intersections. transp res rec. 2001; 1746: 6-13.