Determine Patterns of Causal Factors of Truck Crash Injury Severity on Urban and Rural Roadways Using Fuzzy Cluster Analysis

Abstract

Truck-involved crashes on rural and urban roadways are critical issues and in need of thorough investigations for the crash severity levels. Traditional statistical and machine learning models find difficulty in handling imprecision and producing interpretable models. The objective of this study is to investigate the causal factors influencing truck crash injury severity on rural and urban roadways using Fuzzy Cluster Analysis (FCA). The FCA method was used to assess clustering tendencies, determine optimal cluster numbers, and validate results using Pennsylvania crash data from 2023. The analysis identified 20 clusters across three injury severity categories: Property Damage Only (PDO), Suspected Minor and Possible Injury, and Fatal and Serious Injury. Key findings reveal: (i) urban areas experience rear-end or sideswipe collisions, resulting in PDO or minor injuries (87% of the clusters), while rural areas see more fatal crashes (80% of the clusters); (ii) daylight illumination and licensed drivers are linked to truck crashes across all clusters; (iii) aggressive driving, prevalent in urban settings, correlates with PDO crashes (86% of the clusters); and (iv) clear weather and dry roads are in both PDO and fatal crashes (86% of the clusters). These insights offer actionable pathways to enhance truck safety on diverse roadways.