Safety Performance Function for Motorways using Generalized Estimation Equations

Accident prediction models (APMs) are useful tools for estimating the expected number of crashes over a road network which are typically used in the screening of sites with promise for safety improvements. This study shows a procedure of analysis for motorways network offering a comparison between the conventional analytical techniques based on GLM (Generalized Linear Model) and a different approach based on GEE (General Estimating Equation). The GEE model, incorporating the time trend, is compared in terms of results and reliability in the estimation with conventional models (GLM) that do not take into account the temporal correlation of accident data.