Measuring Nonlinear Relationships and Spatial Heterogeneity of Influencing Factors on Traffic Crash Density Using GeoXAI

Complex relationships require tailored solutions: The study found that factors like road density and neighborhood design don't have simple, linear effects on crash risk. Instead, their impact varies dramatically depending on location—what causes crashes in dense Miami differs from rural areas—meaning one-size-fits-all traffic safety policies are likely ineffective.

Urban density creates distinct safety challenges: Extremely dense urban areas like Miami show "sharply elevated crash risk" primarily due to increased pedestrian activity and complex roadway systems, suggesting that high-density cities need specialized safety strategies rather than applying suburban or rural approaches.

Geography matters more than previously recognized: The research demonstrates "strong spatial heterogeneity" across Florida, with major metropolitan areas (Miami, Orlando, Tampa, Jacksonville) showing significantly different baseline crash risks and factor influences compared to rural regions, highlighting the importance of location-specific interventions.

Socioeconomic factors play a measurable role: Educational attainment was identified as one of the key nonlinear factors influencing crash density, suggesting that traffic safety is connected to broader community characteristics and that equity-focused interventions in disadvantaged areas are warranted.

AI transparency enables better policy design: By using explainable AI methods (GeoShapley), researchers could identify why and where specific factors matter for crashes, enabling policymakers to design targeted interventions like adaptive intersection design and speed management on specific corridors rather than relying on opaque algorithmic predictions.