Spatial analysis of road crash frequency using Bayesian models with Integrated Nested Laplace Approximation (INLA)
DOI: 10.1080/19439962.2020.1726542
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 need for accurate, computationally efficient models to analyze road crash frequency and identify contributing factors, specifically within the context of Indonesia, a region with high crash rates but limited localized research. While Bayesian models are effective for handling complex spatial data, traditional Markov Chain Monte Carlo (MCMC) methods often suffer from long computational times and convergence issues. The authors propose using Integrated Nested Laplace Approximation (INLA) as a faster alternative to MCMC for fitting Bayesian spatial models. The primary objectives are to develop a crash frequency model at the roadway segment level, assess the impact of spatial correlation on model performance, and identify key factors influencing crashes of varying severities. The research focuses on a 225 km highway segment between Banda Aceh and Bireun in Indonesia, divided into 190 segments. The dataset comprises 457 crash records from 2012 to 2015, categorized by severity (fatal, major injury, minor injury, and property damage only), alongside road inventory data including Annual Average Daily Traffic (AADT), land use, and horizontal alignment. The methodology employs a Poisson model with a Besag York Mollié (BYM) specification for spatial effects, implemented via the INLA approach in R. The authors compare spatial models (incorporating Conditional Autoregressive terms) against non-spatial Bayesian models. Model selection and goodness-of-fit were evaluated using the Deviance Information Criterion (DIC), Watanabe-Akaike Information Criterion (WAIC), and Logarithm of the Pseudo Marginal Likelihood (LPML) to cross-validate results and address known limitations of DIC. Results indicate that spatial models consistently outperformed non-spatial models across all crash severity levels. The spatial models exhibited significantly lower DIC and WAIC values and higher LPML values, confirming the importance of accounting for spatial correlation in crash frequency analysis. For instance, the DIC difference for total crashes was nearly 60 points in favor of the spatial model. AADT was identified as the most influential factor across all severity types. However, the impact of land use and horizontal alignment varied depending on the crash severity. Vertical alignment and access density were found insignificant and excluded from final models, while speed limit was omitted due to collinearity with land use. The study concludes that incorporating spatial correlation significantly improves model fit and estimation accuracy for road crash data, particularly at the segment level. Furthermore, the use of INLA provides a robust, computationally efficient alternative to MCMC for Bayesian inference in traffic safety analysis. These findings offer decision-makers in Indonesia and similar contexts a theoretical basis for identifying high-risk segments and implementing targeted safety countermeasures based on specific contributing factors like traffic volume and road geometry.
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 | 2026-06-19 |
| archive | success | semantic_scholar | — | — | 6 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| enrich | success | openalex | — | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| 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