GIS Framework for Hierarchical Bayesian based Crash Data Analysis

A number of steps have been taken over the past several years to advance the Utah Department of Transportation safety initiative. Previous research began the development of a hierarchical Bayesian model to analyze crashes on Utah roadways. The model analyzes roadway segments and determines a posterior predictive distribution of crashes. The actual numbers of crashes for each segment are compared to the predictive distribution and a percentile is calculated. A high percentile indicates more crashes than would be expected and a low percentile indicates less. A Geographic Information System (GIS) framework was developed to facilitate the analysis. The GIS framework has the capability to format the raw data such that it can be read into the statistical model. The GIS framework also displays the numerical data output by the statistical model spatially, allowing for an easy and intuitive analysis of `hot spots' or `black spots.' The purpose of this paper is to outline the GIS framework for crash data analysis, the results of which can be used to further evaluate those segments classified as hot spots.