Probabilistic model for assessing accident rates
DOI: 10.5937/jaes0-42942
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 challenge of efficiently analyzing road accident data to support safety improvement measures. While public statistical databases in Russia provide aggregate accident rates, they lack the granularity to assess specific causes or conditions, such as location type or day of the week, without labor-intensive point-by-point analysis. The authors aim to develop a probabilistic model that converts absolute accident statistics into relative indicators (probabilities), enabling specialists to estimate accident distributions and identify high-risk conditions with minimal processing time. The study utilizes official Russian accident statistics from 2015 to 2021, covering a total of 1,137,987 incidents. The methodology involves analyzing absolute trends in total accidents, urban versus non-urban locations, and daily distributions. Based on this analysis, the authors constructed a probabilistic model where the total probability of an accident $P(A)$ is set to 1. This is decomposed into the probability of an accident occurring in a city $P(A_1)$ and outside a city $P(A_2)$. Similarly, the probability of an accident occurring on a specific day of the week is denoted as $P(B_n)$. The model calculates the joint probability of accidents in specific locations on specific days using the formulas $P(C_n) = P(A_1) \cdot P(B_n)$ for urban areas and $P(D_n) = P(A_2) \cdot P(B_n)$ for non-urban areas. An algorithm was developed to apply these dependencies for estimating incident counts. The results indicate a systematic decrease in total accident rates from 2015 to 2021. The probabilistic analysis revealed that over 70% of accidents occur in urban settlements, with $P(A_1)$ ranging from 0.70 to 0.78 over the study period. Regarding daily distribution, the probability of accidents is relatively stable, with Monday through Thursday and Sunday showing a probability of approximately 14%, while Friday and Saturday show a slightly higher probability of 15–16%. To validate the model, the authors applied it to 2022 data, where the total number of accidents was 126,705. The model’s estimates were compared against actual recorded numbers for each day of the week. The validation showed a maximum relative error of 2.42% (on Wednesday) and a minimum of 0.19% (on Tuesday), with all relative errors remaining below 0.023. The significance of this work lies in providing a streamlined, accurate tool for road safety specialists. By converting complex absolute data into stable probabilistic dependencies, the model allows for rapid assessment of accident-prone conditions without exhaustive manual data processing. The authors conclude that this approach facilitates the targeted implementation of safety measures, such as increased control on specific road sections or days, thereby supporting broader goals like the "Vision Zero" initiative to reduce road fatalities.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-18 |
| archive | success | canonical_url | — | — | 1 | 2026-06-25 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-18 |
| chunk | success | chunk | — | — | 1 | 2026-06-18 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-18 |
| promote | success | — | — | — | 1 | 2026-06-18 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-18 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; 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