Risk-Aware Fleet Management in Public Enterprises: A Machine Learning Approach Using Web-Scraped Data

In recent years, technological advancements, particularly in artificial intelligence and machine learning, have enabled the automation of tasks once thought impractical. However, many public sector organizations continue to rely on manual processes, especially in areas like vehicle fleet management, where critical operational data remains underutilized. This study addresses that gap by proposing a machine learning model aimed at improving vehicle fleet management in public enterprises. The model focuses on classifying drivers based on their risk levels, leveraging behavioral data, individual driver characteristics, and patterns of vehicle usage to provide actionable recommendations for fleet optimization. A key innovation of this work is the integration of web scraping techniques to automatically collect and update data related to drivers, vehicles, and fleet operations. This significantly reduces the dependency on manual data entry and supports the automation of processes such as vehicle registration validity control. The proposed system also includes the development of driver risk classification models, with results visualized through an interactive dashboard and geospatial map to facilitate strategic decision-making. This approach enhances the efficiency, transparency, and data-driven decision-making capabilities of public entities managing transportation assets. © 2025 IEEE.