Understanding Motorcycle Crash Severity in Pakistan: Insights from Machine Learning and Interaction Effects

• Motorcyclists are at a greater risk of sustaining severe injuries or fatalities in road traffic crashes • The study utilized data from a nationally representative department, Rescue 1122, in Pakistan, covering the period from July 2020 and June 2023. • New Jersey barriers, crashes during off-peak hours, rider distraction, cloudy weather, and younger rider age are strongly associated with increased injury severity in these crashes. Motorcycle crashes are a leading cause of road traffic fatalities worldwide, particularly in low- and middle-income countries like Pakistan, where they represent a significant share of road deaths. However, existing research in Pakistan has often relied on self-reported or hospital-based data, typically characterized by small sample sizes, single-year analyses, and a limited focus on roadway geometric features. These limitations have resulted in an incomplete understanding of the factors influencing crash severity. This study aims to address these gaps by analyzing a comprehensive dataset of 15,557 motorcycle crash records from Rawalpindi, Pakistan, spanning the period from July 2020 to June 2023. The dataset includes detailed information on rider demographics, crash causes, vehicle types, weather conditions, and roadway geometry. To predict injury severity, various machine learning models were employed, including Logistic Regression, K-Nearest Neighbors, Decision Tree, Support Vector Machine, Random Forest, Extreme Gradient Boosting, Categorical Boosting. Among these models, the Decision Tree classifier demonstrated the best performance based on precision, recall, F 1 -score, and accuracy metrics. To interpret the model outputs, both global and local SHAP analyses were conducted. The results identified several key factors associated with increased injury severity, including the presence of New Jersey barriers, crashes during off-peak hours, rider distraction, cloudy weather, and younger rider age. In addition, interaction effect plots revealed that speeding significantly increase crash severity, particularly when New Jersey barriers are involved and among younger motorcyclists. In light of study findings, several policy measures are proposed, including educational and awareness initiatives for young riders, strategies to regulate mixed traffic, and improvements to infrastructure aimed at enhancing motorcyclist safety in Pakistan and other regions facing similar road safety challenges.