Dr. Şebnem Er (Statistical Sciences Department, University of Cape Town, South Africa): "A Bayesian mixture model for monitoring traffic accident trends".

Data i lloc: Friday, December 12, 2025 · 12:00 h · Seminar of the Departament d'Estadística i Investigació Operativa, of the Facultat de Ciències Matemàtiques.

Abstract: In this talk, we describe a multinomial spatiotemporal Bayesian mixture model to characterize the distribution of monthly accident severity outcomes (fatal, serious, minor, and no injury) at police stations in the city of Cape Town between January 2015 and October 2023. The model jointly estimates spatially varying accident type probabilities, captures temporal evolution through structured and unstructured random effects, and allows for probabilistic detection of outliers at the station level. In addition, the model allows for the identification of stations that exhibit overall increasing, decreasing, or stable trends for certain accident types.

•