Large‐scale road safety evaluation using extreme value theory
DOI: 10.1049/iet-its.2019.0633
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the limitations of traditional road safety analysis, which relies on crash data that is often sparse, unreliable, and reactive. To overcome these issues, the authors apply Extreme Value Theory (EVT) to surrogate measures of safety, specifically Time-to-Collision (TTC), to predict the probability of rear-end collisions on motorways. The study aims to demonstrate the feasibility of EVT for large-scale, real-world applications by using long-duration data collection and automated model calibration, filling gaps in previous research that typically relied on short observation periods and subjective threshold selection. The methodology involves a large-scale case study on two motorways in north-eastern Italy. Vehicle-by-vehicle traffic data were collected over one year (2013) at 19 cross-sections using microwave Doppler radars. TTC was calculated for consecutive vehicles and normalized to account for variability. The authors applied two EVT approaches: Block Maxima (BM), which uses daily maximums of negated TTC values to fit a Generalized Extreme Value distribution, and Peak-Over-Threshold (POT), which fits a Generalized Pareto distribution to values exceeding a specific threshold. To address the computational burden of manual threshold selection in POT, the authors implemented an Automatic Threshold Selection Method (ATSM). Model predictions for annual rear-end collisions were validated against a six-year historical crash database (2011–2016), which recorded 5,632 total crashes, with rear-end collisions accounting for 29% of incidents and 46% of fatal crashes. The results indicate that both BM and POT approaches produced reliable predictions. In approximately 90% of the cross-sections, the observed number of annual rear-end crashes fell within the 95% confidence interval of the predicted values. The BM approach generally yielded lower predicted collision counts compared to POT, though both methods successfully captured the risk profile of the sections. The use of normalized TTC and the ATSM allowed for efficient, automated calibration across all 19 sections without significant loss of accuracy. The study confirms that rear-end collisions are a critical safety concern on these motorways, justifying the focus on this specific crash type. The significance of this work lies in its demonstration that EVT can be effectively applied to large-scale, long-duration traffic data using cost-effective radar technology. By introducing an automated threshold selection method, the authors reduce the subjectivity and time requirements associated with POT applications, making EVT more practical for widespread road safety evaluation. This approach enables proactive safety assessment using surrogate measures, allowing for quicker identification of high-risk locations compared to waiting for sufficient crash data to accumulate. The findings support the integration of EVT into intelligent transport systems for continuous, data-driven road safety monitoring.
Key finding
Extreme value theory applied to radar-collected time-to-collision data reliably predicted annual rear-end collisions, with observed crashes falling within the 95% confidence interval of predictions for about 90% of the evaluated motorway cross-sections.
Methodology
on_road
Sample size: 19
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | openalex | — | — | 9 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | — | — | — | 1 | 2026-05-28 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: crash risk outcomes