Spatio-temporal heterogeneity in street illegal parking: A case study in New York

Illegal parking has a significant impact on urban traffic management and safety, posing a substantial hazard that contributes to disorder in public urban spaces. Therefore, a thorough analysis of the temporal and spatial characteristics of illegal parking is essential for the scientific planning of parking areas and the optimization of traffic resource allocation . However, existing studies often oversimplify factor interactions and fail to disentangle spatio-temporal heterogeneity, particularly in addressing zero-inflated data structures and nonlinear dependencies among variables such as crime rates and weather conditions. To this end, this study constructed a multi-spatiotemporal scale Bayesian hierarchical model that combines the Besag-York-Mollié (BYM) model with a zero-inflated Poisson distribution and uses the Integrated Nested Laplace Approximation (INLA) method for efficient posterior inference. In contrast to conventional approaches, this method not only improves interpretability but also precisely captures spatial-temporal dependencies, thereby enabling a more nuanced and holistic characterization of parking violation dynamics. The results show that illegal parking has obvious spatiotemporal heterogeneity; road and population density are the main drivers of illegal parking, exacerbating the imbalance between supply and demand; crime rate and traffic demand amplify illegal behaviors in the central city, while transportation infrastructure suppresses risks by promoting the use of public transportation; humidity has the greatest impact on parking behavior, exceeding the effects of temperature and visibility. SHAP analysis further reveals nonlinear interactions, indicating that crime rate dominates risk prediction when combined with variables such as traffic demand. The research findings offer valuable decision-making insights for optimizing the urban traffic management system and enhancing targeted parking policies. • Spatio-temporal modeling effectively forecasts illegal parking locations and frequencies in urban settings. • A Bayesian hierarchical spatio-temporal model integrates multi-source data, improving interpretability and reducing computational cost. • Key factors influencing illegal parking include road and population density, crime rate, traffic demand, and weather conditions. • Weather impacts illegal parking, with humidity showing the most significant positive effect.