Abstract

Driving behavior analysis is essential for enhancing traffic safety and informing insurance practices. Research highlights the significance of understanding consumer driving habits to refine risk assessments. Additionally, insights from driving data can improve user confidence in vehicle safety technologies. However, many developers struggle to effectively analyze and visualize this complex data. This paper seeks to answer the question: “How can driving behaviors be quantitatively assessed and visualized?” We propose the DIVAS visualization system, leveraging a UBI-based scoring model and unsupervised learning to evaluate driving events using data from the 2023 ChinaVis Data Challenge.