A0478
Title: Spatial modelling road accidents in Besancon (France) using log-gaussian cox processes
Authors: Cecile Spychala - Universite de Franche-Comte - Lmb (France) [presenting]
Camelia Goga - CNRS-LMB (France)
Clement Dombry - Universite de Franche Comte (France)
Abstract: In order to prevent and/or forecast road accidents, the statistical modelling of spatial dependence and potential risk factors is a major asset. The focus is on the georeferenced location of accidents. We crossed these events with covariates characterizing the study geographical area such as sociodemographic and infrastructure measures. After a variable selection (Poisson model, Poisson models aggregation and random forest), the occurrence of accidents was modelled by using a spatial log-Gaussian Cox process. The results of this analysis enable us to identify principal risk factors for road accidents and critical areas. The data used are road accidents that occurred between 2017 and 2019 in the CAGB (urban community of Besancon).
