Comparative Analysis of Machine Learning Algorithms for Traffic Flow Prediction in Intelligent Transportation Systems

As human society advances, the limitations of current transportation systems become more apparent, prompting a growing desire to leverage advanced technology for intelligent transportation solutions. However, low identification rates and complicated processing plague many of the vehicle detection techniques used in video analytics today. This paper gives a comparative analysis of multiple machine learning algorithms for traffic flow prediction in order to improve Intelligent Transportation Systems (ITS). Using an extensive dataset of historical traffic flow and environmental variables, we evaluate the performance of several methods: Linear Regression (LR), Support Vector Machine (SVM), Decision Trees, Random Forests (RF) and deep Neural Networks (DNN). Numerous Variables are taken into account in the evaluation, including scalability, computing efficiency, and forecast accuracy. The findings show that while algorithms like RF and SVM give good performance with less computational overhead, DNN obtain the maximum accuracy. The work offers valuable information on the disadvantages of various machine learning (ML) techniques, which may be used to select suitable models for the prediction of traffic movement in real-world ITS.