The increase of congestion in the urban environment has become a major problem. The traditional traffic control methods with poor administration of human resources do not moderate traffic, resulting in increased traffic congestion and road violations. The intelligent transportation system (ITS) can guarantee safety, efficiency, and sustainability for large-scale vehicle traffic issues. In order to ensure the smooth flow of traffic, ITS combines machine learning with the existing traffic control system and provides a real-time strategy. Several researchers have shown great work with different optimization techniques in intelligent traffic police management and deployment. However, it remains necessary to compile such an impressive effort as a whole. In light of these facts, we present a comprehensive review of the state-of-the-art technology for the development of a three-tier solution classification in machine learning. The first tier contains several tools and methods for collecting traffic statistics. The second tier focuses on the accuracy of the machine learning algorithms, forming a pattern for the acquired data, and then provides important data on traffic flow, congestion levels, and so on. Various traffic planning techniques are covered in the third tier, the most essential layer of taxonomy. The proposed review also examines the usage of traffic police schedules which develop the application of this evaluation in different areas. Finally, some of the major challenges are discussed and further improvement is initiated.