Traffic modeling using Machine Learning methods for predicting vehicle numbers at junctions: A case study in Colombo, Sri Lanka (preprint)

ABSTRACT

Traffic modeling is a fundamental tool for comprehending the complex dynamics of urban mobility. It provides vital insights for transportation planning and the construction of infrastructure. This study aims to employ sophisticated machine-learning techniques to forecast the volume of vehicles at certain main intersections in Colombo, Sri Lanka. The study emphasizes the importance of traffic modeling in influencing policy development and optimizing transportation networks, based on a thorough evaluation of relevant literature. The research utilizes machine learning methods, namely random forest regression, to analyze temporal and spatial trends in traffic flows. This analysis yields valuable insights for urban planners and transportation authorities. An examination of the dataset, utilizing methodologies such as rolling statistics and time series analysis, reveals subtle variations in traffic levels over time, including the influence of external influences like as the COVID-19 pandemic. The study's results highlight the significant impact of incorporating machine learning techniques into traffic modeling, providing a solution to improve urban mobility, decrease congestion, and create more sustainable transportation systems. This research ultimately enhances the development of data-driven solutions to tackle the changing mobility requirements of Colombo and establishes the groundwork for future research in the field of urban transportation modeling.