Macro level Safety Analysis in Urban Using Conditional Autoregressive Model

Abstract:

This study collected crash data, roadway features, traffic characteristics and land use data in 263 traffic analysis zones (TAZ) within the Outer Ring in Shanghai. TAZ level Bayesian Negative Binomial Conditional Autoregression model was established. Modeling results show that the frequency of total crashes in each TAZ can increase with longer arterials and minor arterials. The numbers of intersections and road density are positively correlated with crashes. More crashes occur with the increasing of car production and land development intensity.

Key words:macro level safety analysis; traffic analysis zone(TAZ); Bayesian estimation; conditional autoregressive model; risk factor