Differential Analysis Discerned Motorcycle Risk Factors on Traffic Control and Law in Developing Country’s Urban Context

As a developing country, Bangladesh is burdened by the rising number of motorcycles on the roads, which make up most vehicles and cause the bulk of traffic accidents. To address this critical issue, this paper aims to address the lack of comprehensive research on motorcycle accidents by exploring the impact of the perception of motorcycle crashes on traffic control and law enforcement. Prior studies have concentrated on real-time crash prediction and identifying significant contributors to crashes, neglecting simultaneous consideration of factors like driving environment, weather conditions, traffic control, driving behavior, and pedestrian-related characteristics. To bridge this gap, the study employs advanced machine learning algorithms, namely the Random Forest and CNN 1D algorithm, to analyze distinctive responses from various types of road users: motorcycle riders, motorcycle users (females and males) with different graphical illustrations. This research endeavors to establish a robust model of precursors by gathering ratings from road users, enabling effective design, management, planning, and implementation of policies to improve traffic control and law enforcement to reduce motorcycle accidents. ‘Traffic movement,’ ‘sign marking and lighting,’ and ‘driving environment’ related features were proved to be the most significant precursors. Furthermore, the study has deployed the machine learning model on a public server, providing policymakers and users with a user-friendly interface to predict the target variable ‘traffic control and law’ based on their input. Users can enter ratings on relevant risk factors using a 1 to 5 scale to generate predictions.