Investigation of the Hourly and Spatial Patterns of Traffic Offenses During March–April 2019 in Iran Using Bivariate Generalized Additive Models and Integrated Nested Laplace Approximation
Published in International Journal of High Risk Behaviors and Addiction, 2022
Title
Investigation of the Hourly and Spatial Patterns of Traffic Offenses During March–April 2019 in Iran Using Bivariate Generalized Additive Models and Integrated Nested Laplace Approximation
Authors
Mohammad Fayaz, Alireza Abadi, Alireza Razzaghi, Soheila Khodakarim, Mostafa Hosseini
Journal Information
Journal: International Journal of High Risk Behaviors and Addiction
Year: 2022
Volume: 11
Issue: 3
Article: e118376
DOI: https://doi.org/10.5812/ijhrba-118376
Abstract
Background:
Traffic offenses, including speeding and unsafe following distance, represent important indicators of risky driving behavior and road safety risks. High-resolution traffic monitoring systems allow detailed spatiotemporal modeling of such behaviors.
Objectives:
This study investigates hourly and spatial patterns of traffic offenses in Iran using generalized additive models (GAMs) and integrated nested Laplace approximation (INLA).
Methods:
Data from 2,316 traffic count stations collected during March–April 2019 were analyzed. Hourly averages of speeding and unsafe distance offenses were computed. Correlation analyses were performed using Pearson and Spearman coefficients. Several statistical distributions were evaluated within a GAM framework, and model performance was assessed using mean squared error (MSE) and correlation measures.
Results:
The bivariate Gaussian GAM demonstrated the best predictive performance. Three distinct hourly patterns of traffic offenses were identified, reflecting varying dominance between speeding and unsafe distance violations. Spatial analysis revealed heterogeneous regional risk patterns across Iran, with higher speeding rates in central, northeastern, and southeastern regions, and higher unsafe distance violations in northern, northwestern, western, and southwestern regions.
Conclusion:
Traffic offenses in Iran exhibit complex spatiotemporal dynamics. The findings highlight distinct regional risk structures and emphasize the importance of advanced statistical modeling for traffic safety analysis.
Full Text
Supplementary Material
Additional supporting analyses and extended results are available as supplementary material:
Keywords
- Traffic Offenses
- Generalized Additive Models
- INLA
- Spatiotemporal Analysis
- Road Safety
- Iran
- Functional Data Analysis
Citation
Fayaz M, Abadi A, Razzaghi A, Khodakarim S, Hosseini M. Investigation of the hourly and spatial patterns of traffic offenses during March–April 2019 in Iran using bivariate generalized additive models and integrated nested Laplace approximation. Int J High Risk Behav Addict. 2022;11(3):e118376.
Recommended citation: Fayaz M, Abadi A, Razzaghi A, Khodakarim S, Hosseini M. Investigation of the hourly and spatial patterns of traffic offenses during March–April 2019 in Iran using bivariate generalized additive models and integrated nested Laplace approximation. Int J High Risk Behav Addict. 2022;11(3):e118376.
Download Paper
