چكيده به لاتين
The characteristics of the road network have a special effect on the safety of the roads so that the type of topology, geometry, and configuration of the road network can change the probability of traffic accidents. In order to investigate the spatial effects of the network, the network has been quantified using angular space syntax methods. The purpose of this study is to investigate the effect of space syntax parameters in the construction of spatial models of the accidents frequency. In this study, for the first time, the method of segment analysis or angular analysis, which also has angular effects in quantification, has been used in the construction of Poisson and negative binomenal geographic weighted regrresion models in the analysis of accident frequency in segment level. Negative binomial models have emerged to cover the effect of data overdispersion. The urban network of Mashhad has been selected to space syntax anlysis of and the parameters of angular connecitvity, integration and choice for main roads and highways in the model have been extracted. The average frequency of urban accidents in Mashhad in the years 85, 86 and 87 for each segment is considered as a dependent variable. Exposure variables in three groups of AADT and road length, vehicle kilometer traveled and total delay time have entered in models. Other descriptive variables used to participate in the models include speed, type of road (secondary, main and highway), road width and one-way road. Among the various modes of modeling, the negative binomial macro model of geographical weight regression provides the best crash estimation. The reason for using geographical regression models is to consider spatial heterogeneity in estimating descriptive variables. In general, macro and local inetgration within a radius of 400 m, angular connectivity, natural logarithm of path length and ADDT, road width and one-way path are the main model variables. Among these descriptive variables, on average, except for angular connectivity and local inegration, the other variables increase the frequency of urban accidents. It can be said that the more the integration of the road increases the accidents freqeunsy. Increasing the local integration parameter indicates that commercial areas, which is facing a reduction in accidents. Higher angular connectivity of the segments increases the complexity of network, which will make drivers more cautious. Among the exposure variables, road length and AADT have provided the best performance among the group of exposure variables.
