Intelligent Transportation System's Machine Learning-Based Traffic Prediction

Abstract

The aim of this study is to develop an accurate and timely traffic flow prediction tool that considers various factors influencing road conditions, such as road repairs, rallies, traffic signals, and other everyday events that can impact traffic movement. By providing drivers with near real-time predictive insights, they can make more informed decisions, enhancing traffic management and potentially supporting future autonomous vehicle technologies. Given the exponential growth in traffic data, this research applies big data principles to the transportation domain, where existing traffic prediction models struggle to handle real-world applications effectively. In this study, we implemented machine learning, genetic algorithms, soft computing, and deep learning techniques, achieving a traffic flow prediction accuracy of 93.5%. The results demonstrate a significant improvement in prediction accuracy compared to conventional models, which typically average around 85%. Additionally, image processing algorithms for traffic sign identification are integrated, achieving 90% accuracy in identifying key traffic signs, further aiding in the training of autonomous vehicles. The proposed approach addresses the challenges posed by large-scale transportation data, offering a solution with improved predictive accuracy and practical utility.