Evaluating the spatial effects of environmental influencing factors on the frequency of urban crashes using the spatial Bayes method based on Euclidean distance and contiguity

Abstract

Built environmental factors are one of the most important causes of urban crashes. Studies have shown that in addition to crash data, which have spatial heterogeneity, factors influencing crashes also have a spatial correlation. The main goal of this study is to evaluate the spatial effects of environmental factors on the frequency of crashes in Shiraz, Iran, at the TAZ level. In the first step of the study, using component analysis models, important environmental factors affecting the crash were identified, and composite indicators were produced as independent variables. In the second step, to control the effect of correlation and heterogeneity of model variables, spatial statistical models based on Euclidean distance such as geographically weighted Poisson regression (GWPR), geographically weighted negative binomial distribution (GWNBR), as well as Poisson and distribution models Negative binomial based on neighbor distance is used in spatial Bayes method with INLA approach. The study's results showed that models based on distance and contiguity to evaluate the spatial effects of crash data and the factors affecting it at the TAZ level have higher accuracy than geographically weighted regression models, as well as indicators of land use diversity and access to the system. The public transport produced in the first step effectively increases the frequency of crashes, and in TAZs where this index is high, there is a higher probability of a crash. The results of this study can be important for city managers and planners to improve urban safety measures, development planning, and future city measures.