AN ARTIFICIAL INTELLIGENT APPROACH TO TRAFFIC ACCIDENT ESTIMATION: MODEL DEVELOPMENT AND APPLICATION
DOI: 10.3846/1648-4142.2009.24.135-142
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 public health and socio-economic issue of traffic accidents in Turkey, specifically focusing on Ankara, where vehicle ownership has surged significantly compared to population growth. The research aims to develop robust models for estimating the number of accidents, fatalities, and injuries to assist decision-makers in formulating future road safety strategies. To achieve this, the authors compare two artificial intelligence approaches: Artificial Neural Networks (ANN) and Genetic Algorithms (GA). The study utilizes historical data from 1986 to 2005, selecting population and the number of vehicles as input parameters due to their availability and relevance, while excluding more complex factors like driver behavior or road geometry. The methodology involves training and testing both ANN and GA models. The ANN model employs a feed-forward back-propagation algorithm with a 2-5-1 network architecture, using sigmoid and linear activation functions. Data from 1986–2000 served as the training set, while 2001–2005 data was used for testing. The GA approach developed two forms of mathematical expressions—linear and exponential—for accident, injury, and fatality prediction. GA parameters included a population size of 55, 200 generations, and specific crossover and mutation probabilities. Model performance was evaluated using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The results indicate that the ANN model significantly outperformed the GA model in terms of accuracy. For the GA models, the exponential form proved superior for estimating accidents and fatalities, while the linear form was more appropriate for predicting injuries. However, the ANN models consistently yielded lower MAE and RMSE values across all categories during both training and testing periods. For instance, in the testing period, the ANN’s MAE for accident estimation was 1,973, compared to the GA’s 4,049. Consequently, the ANN model was selected for future projections. The study applied the ANN model to two scenarios for the period 2006–2020: one assuming historical growth rates for population (2.0%) and vehicles (7.5%), and another assuming vehicle ownership reaches 60% of the population. Both scenarios predicted a rise in accidents and injuries but a decrease in fatalities by 2020. Validation against actual 2006–2007 data showed that observed accident and injury numbers fell within the estimated ranges, though fatality estimates were less accurate, likely due to unmodeled variables like seat belt usage. The significance of this research lies in demonstrating the utility of artificial intelligence, particularly ANN, for traffic safety planning in regions with rapid motorization. By providing reliable estimates of accident trends under different growth scenarios, the study offers a practical tool for policymakers to anticipate safety challenges and implement targeted interventions. The findings suggest that while vehicle numbers may increase, strategic planning informed by such models can help mitigate the severity of traffic outcomes, highlighting the potential of AI in transportation engineering.
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 | 2026-06-25 |
| archive | success | canonical_url | — | — | 1 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-25 |
| chunk | success | chunk | — | — | 1 | 2026-06-25 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-25 |
| promote | success | — | — | — | 1 | 2026-06-25 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-25 |
| 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
- Theoretical Contribution: computational model