Risk Factor Analysis for Signalized Intersections Along Corridors with a Consideration of Spatial Correlation

Abstract:

Generalized estimating equation (GEE) was applied to account for the correlation among intersections from corridors. Risk factors including intersection location, geometric design, signal control, and operation condition were identified. The results show that GEE with the autoregressive correlation structure is the most effective way to deal with the spatially correlated data. It confirms that intersections in close proximity, with greater ratio of turning lanes, more number of phrases, and more ADT tend to have higher crash frequencies; 4 legged intersections and intersections under elevated expressways are associated with more crashes.