Hit and run crashes: Knowledge extraction from bicycle involved crashes using first and frugal tree
DOI: 10.1016/j.ijtst.2018.11.001
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study addresses the critical safety issue of hit-and-run crashes involving bicyclists, a vulnerable road user group for whom such incidents often result in severe injuries or fatalities due to delayed emergency assistance. While hit-and-run behavior is a punishable offense, it remains frequent, and existing research has largely overlooked the specific environmental and crash characteristics that trigger drivers to flee scenes involving bicycles. The authors aim to identify key factors contributing to bicycle-involved hit-and-run crashes to inform safety planning and operations, filling a gap in literature that has predominantly focused on general hit-and-run statistics or used complex, less interpretable models. The researchers utilized six years (2010–2015) of police-reported crash data from Louisiana, comprising approximately 108,803 hit-and-run crashes, of which over 1,000 involved bicycles. The dataset was merged from crash, roadway inventory, vehicle, and occupant tables to isolate bicycle-involved incidents based on specific vehicle type and harmful event indicators. To analyze the data, the authors applied the Fast and Frugal Tree (FFT) heuristics algorithm, specifically the fan algorithm variant, which is designed for binary decision-making. This method was chosen for its simplicity, robustness against overfitting, and ability to handle noisy data better than complex machine learning models like support vector machines or random forests. The dataset was split into training and testing subsets to develop and evaluate the predictive model’s performance. The analysis revealed that bicycle-involved hit-and-run crashes accounted for 22% of all bicycle crashes in Louisiana during the study period, with fatal bicycle crashes representing 10% of all fatal hit-and-run incidents. The FFT model identified five major cues significantly contributing to the prediction of bicycle involvement in hit-and-run crashes: crash severity (fatal and injury crashes), collision type (right-angle, turning, or head-on), location type (city streets or other urban areas), intersection presence, and locality type (residential or mixed). The model achieved a balanced accuracy of approximately 76% for both training and testing data. Notably, the FFT model demonstrated higher sensitivity compared to other complex, black-box machine learning approaches, providing clearer insights into the decision rules governing these crashes. The findings underscore the importance of specific roadway environments and crash characteristics in predicting hit-and-run behavior involving bicyclists. By identifying that crashes in urban intersections, residential areas, and those involving severe injuries or specific collision angles are more likely to involve bicycles in hit-and-run scenarios, the study offers actionable insights for transportation planners. The use of FFT highlights a viable, interpretable alternative to traditional statistical and complex machine learning methods for traffic safety analysis, enabling more transparent decision-making processes for crash reduction strategies.
Key finding
The fast and frugal tree model identified fatal/injury status, collision type, and roadway location as the primary predictors of bicycle-involved hit-and-run crashes, achieving 76% balanced accuracy.
Methodology
modeling
Sample size: 108803
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 | canonical_url | — | — | 11 | 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