Dr. Şebnem Er (Statistical Sciences Department, University of Cape Town, South Africa): "Spatio-Temporal Zero-Inflated Negative Binomial Modeling of Traffic Accidents at Road Intersections in Cape Town, South Africa"

Data i lloc: Dijous 12 de desembre de 2024 · 12:00 h · Aula 1.7 de la Facultat de Ciències Matemàtiques

Resum: In this research, yearly aggregated road traffic accident counts at road intersections in one of the busiest wards in Cape Town between 2015-2019 are analysed using spatio-temporal zero-inflated negative binomial models through a Bayesian approach. Road level information on speed limits, speed bumps, traffic lights, road class, number of lanes, if an intersection is on "Main Road, Cape Town", and if a minibus taxi or MyCity public bus route passes through the intersection are used as the fixed effects covariates for modeling the number of road traffic accidents. Results reveal that amongst the covariates included in the final model, degree, the existence of traffic signals, road class, existence of main road and a taxi route at the intersection, and finally the dummy variables for the years 2018 and 2019 all increase the traffic accident counts at the intersections. For the zero-inflation part of the model, the existence of traffic signals and an intersection being on main road both have negative coefficients indicating that these covariates reduce the odds of being a false zero accident whereas speed increases the odds.