An analysis of existing traffic models is presented, and the problems of processing the data used are revealed. The evolution of approaches to traffic management is considered, from classical analytical models to modern artificial intelligence methods. Special emphasis is placed on the key role of big data as a source of information for training, calibration, and operation of these models. The challenges and prospects of this subject area are described. A classification of road traffic models is given, micro-, macro-, and mesoscopic models are described, their key features, advantages and disadvantages are listed, and a mathematical apparatus describing them is formed. It is shown that the effectiveness of any control algorithm directly depends on the volume, quality and depth of analysis of incoming data. The methods of traffic modeling are systematized and their relationship with big data analysis technologies is demonstrated. The application of machine learning in traffic management tasks is described. The scientific novelty of the fundamental traffic model proposed by the authors is substantiated — this macroscopic model has the advantage of calculation speed, ease of modeling the traffic situation, and can be used to evaluate the regulated parameters of the specified characteristics of the road infrastructure.