A new data-driven model for vehicle and pedestrian safety: statistical approach based on spatial decision-making

Area of Science:

• Traffic Safety and Accident Analysis
• Spatial Statistics
• Risk Factor Identification

Background:

• Minimizing traffic accident losses is crucial for public safety.
• Identifying and analyzing risk factors is essential for accident severity reduction.
• Previous studies often lack integrated spatial and statistical approaches to risk assessment.

Purpose of the Study:

• To propose and validate a novel spatial decision-making-based statistical methodology for determining traffic accident risk factors.
• To analyze risk factors across three distinct collision types: vehicle-vehicle, vehicle-pedestrian, and vehicle-other.
• To provide actionable insights for evidence-based traffic safety policy development.

Main Methods:

• Utilized 5-year (2015-2019) accident data, defining 22 independent and 157 sub-variables.
• Employed the fuzzy simple weight calculation method to assess the influence of risk factors.
• Integrated spatial analysis via geographical information system (GIS) with multinomial logistic regression.
• Validated the model with a McFadden R² of 0.749, indicating a strong fit.

Main Results:

• Identified significant variables increasing or decreasing the probability of each crash type.
• Found geo-intersections to be the highest risk factor for vehicle-vehicle crashes.
• Determined pedestrian defect as the primary impact factor for vehicle-pedestrian crashes.
• Spatial analysis revealed higher accident severity in western, southern, and central regions of Türkiye.

Conclusions:

• The proposed methodology provides a comprehensive framework for understanding and mitigating traffic accident risks.
• Findings offer critical guidance for policymakers and traffic safety experts to enhance vehicle and pedestrian safety.
• Results underscore the importance of spatial context and specific contributing factors in accident severity.