Identifying accident and safety risks in urban micromobility: An analytical approach using machine learning

Micromobility has recently emerged as a popular and sustainable mode of transportation in urban areas. Understanding the accident-related factors of this new transport mode is crucial to ensure safe and enjoyable user experience. This study applies a machine learning approach, specifically the decision tree algorithm, to investigate those factors that lead to accidents by drivers using three types of micromobility devices: normal bicycle, e-bicycle, and e-scooter. This method can determine the relationship between various attributes and target variables. The research methodology adopted in this study includes a survey analysis involving questionnaires conducted among residents of Budapest, Hungary. The survey covers various aspects such as socio-demographics, mobility patterns, trip types, riding locations, and safety concerns. The attributes are analyzed with the elaborated method using the Python programming language. The results of the study demonstrate that the attitude of frequently wearing a helmet and the trip type are the primary elements related to accidents in the process. Moreover, variables such as age, gender, and riding location contribute to accident occurrence, as well. Policy implications include promoting helmet usage and enhancing the infrastructure to improve micromobility safety.