Co-occurrence patterns of risk factors for fatal pedestrian crashes on rural corridors: An association rule mining approach

Introduction Pedestrian fatalities on high-speed rural roads in India continue to pose a persistent safety challenge. This study uses the Association Rule Mining (ARM) approach to systematically uncover multidimensional interaction patterns among roadway geometry, traffic, landuse, pedestrian volumes, demographic, and roadside attributes associated with pedestrian-involved fatal crashes along the rural corridors. Methods All variables were discretized into distinct categories and transformed into segment-level transactional itemsets. Using the Apriori algorithm, association rules were mined with a fixed consequent (pedestrian crash in a segment: yes), and optimal support-confidence thresholds (0.01 and 0.50) were identified through a grid-search sensitivity analysis to ensure both rule quality and interpretability. Results The ARM results reveal consistent, high-lift patterns involving the coexistence of junctions, settlements, and authorized median gaps with several rules having lift values above six. Other influential antecedents include paved shoulders, 4-lane cross-sections, NH-segments, median without barriers, straight alignments, and mid-to-high vehicle operating speeds. Four visualization plots (scatterplot, grouped balloon plot, network graph plot, and parallel coordinate plot) of all the rules illustrate the structural dominance of junction-settlement-median gap triad configurations in rules of both moderate (4-item) and higher (5-item) complexity. Conclusions The findings highlight the heightened risk posed by specific combinations of uncontrolled pedestrian access points, roadside activities, cross-sectional designs, and high operating speeds, offering actionable insights for targeted interventions such as median gap consolidation, junction access control, speed management in settlements, and pedestrian-sensitive geometric redesign. The study demonstrates the ARM's effectiveness as a pattern-oriented analytical approach for assessing pedestrian safety on rural high-speed corridors.