Matin Matin - PhD Candidate of Civil Engineering, School of Civil Engineering, Iran University of Science& Technology
Afshin Shariat Mohaymany - Associate Professor of Civil Engineering, School of Civil Engineering, Iran University ofScience & Technology

The rapid expansion of road construction and ever-increasing growth ofurbanization have led to increased number of vehicles. Achieving the safe tripswithout personal harm or property damage has always been the concern of safetyspecialists. Over the last few years the safety researchers attempted to developthe innovative methodologies to explore the crash affecting factors and obtainpractical models with high prediction power. Generalize Linear Models (GLMs)with negative binomial or Poisson distribution for errors have performedsuccessfully in this case. Such models assume the dependent variable (e.g. crashfrequency or crash rates) to be statistically independent, however; trafficaccidents indicate different degrees of dependency, a phenomenon known asspatial autocorrelation. Values over distance are more or less similar thanexpected for randomly associated observations. This study aims to develop thespatial semi-local crash prediction model using Poisson-Gamma-CAR. Themodel was employed over 253 TAZ in Mashhad, Iran in 2008 to explore therelationship between the crashes and the explanatory variables. The resultsconfirmed that the model performs well to model the over-dispersion of crashdata; either to capture the significant spatial dependencies. The results ofgoodness-of-criteria have also proved the model’s strength in predicting thecrashes over TAZ-level.

Spatial Dependency, Poisson-Gamma-CAR-Safety