A Bayesian spatio-temporal interaction (B-ST-I) approach was developed for hotspot identification in the context of macroscopic safety analysis.
Models accounting for space-time effects perform better in safety ranking than does the Poisson-lognormal model.
The B-ST-I model significantly outperforms the Bayesian spatial and temporal model.
The B-ST-I model provides insight into the processes influencing changing crash rates over time and efficiently improves zonal-level safety.