A Dimensionality-Reduced XAI Framework for Roundabout Crash Severity Insights

2025-09-15

AI Summary

This study addresses the challenge of dynamically varying crash severity at roundabouts under heterogeneous contextual conditions. Leveraging 2017–2021 crash data from Ohio roundabouts, we propose a two-stage interpretable AI framework: (1) clustering correspondence analysis (CCA) to identify four prototypical crash patterns; and (2) integration of tree-based models with SHAP values to enable attributional linkage from group-level patterns to individual crash instances. Our approach is the first to systematically quantify the nonlinear effects of illumination, pavement condition, vehicle speed, and collision type on injury severity, uncovering context-dependent risk pathways. The resulting methodology establishes an auditable, reproducible analytical template for public safety applications, directly supporting evidence-based hotspot identification and context-sensitive safety interventions—thereby enhancing both the scientific rigor and interpretability of roundabout risk governance.

Technology Category

• Reasoning under Uncertainty: Causality
• Computer Vision: Interpretability, Explainability, and Transparency
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Security and Privacy: Data transparency and provenance
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Roundabouts reduce severe crashes, yet risk patterns vary by conditions. This study analyzes 2017-2021 Ohio roundabout crashes using a two-step, explainable workflow. Cluster Correspondence Analysis (CCA) identifies co-occurring factors and yields four crash patterns. A tree-based severity model is then interpreted with SHAP to quantify drivers of injury within and across patterns. Results show higher severity when darkness, wet surfaces, and higher posted speeds coincide with fixed-object or angle events, and lower severity in clear, low-speed settings. Pattern-specific explanations highlight mechanisms at entries (fail-to-yield, gap acceptance), within multi-lane circulation (improper maneuvers), and during slow-downs (rear-end). The workflow links pattern discovery with case-level explanations, supporting site screening, countermeasure selection, and audit-ready reporting. The contribution to Information Systems is a practical template for usable XAI in public safety analytics.

Problem

• Analyzing roundabout crash severity patterns using explainable AI
• Identifying co-occurring risk factors through cluster correspondence analysis
• Quantifying injury drivers with SHAP interpretation across crash patterns

Innovation

• Cluster Correspondence Analysis identifies crash patterns
• Tree-based model interpreted with SHAP for severity
• Workflow links pattern discovery with case explanations