Spatial Analysis of Road Accidents on the Czech D1 Motorway: Hotspots, Traffic Exposure and Road Geometry
This paper presents a spatial analysis of road accidents on the Czech D1 motorway, covering the Prague–Říkovice and border–Přerov sections in both directions. The data comprise 72,141 road segments of about 10 m (721.5 km of carriageway) with 876 recorded accidents, traffic volumes (AADT by vehicle class), lane and width attributes and a bend angle. Segments were aggregated to 722 one-kilometre sections.Three research questions are addressed: (1) where the accident hotspots are and whether they are statistically significant, (2) how traffic volume and heavy-vehicle share relate to accident counts, and (3) whether road geometry helps explain or predict accident locations.Methods: Getis-Ord Gi* hotspot analysis and Moran’s I with permutation inference; negative binomial regression with a length offset; and a random forest classifier evaluated with 5 km spatial block cross-validation.Main results: Accidents are significantly clustered on all four carriageways (Moran’s I = 0.10–0.49). Seventeen sections are significant hotspots. The eight that fall among the fifteen most accident-heavy sections all lie roughly 11–20 km from Prague, on both carriageways. Traffic volume is the strongest correlate of accident counts (IRR for log AADT = 5.07). Sharp bends (IRR = 1.07) and carriageway width (IRR = 1.16) have weaker but significant associations, and heavy-vehicle share is marginal (p = 0.061). Predictive power is modest (mean ROC-AUC 0.665, mean PR-AUC 0.021 against a baseline of 0.011).
Authors
- Sareena Bilal
Publication Details
- Journal
- Figshare
- Published
- 2026-11-15
- DOI
- https://doi.org/10.6084/m9.figshare.35013406
- Primary Topic
- Traffic and Road Safety
- Type
- preprint