Urban traffic flow prediction techniques: A review

• Published in: Sustainable computing informatics and systems Vol. 35; p. 100739
• Main Authors: Medina-Salgado, Boris, Sánchez-DelaCruz, Eddy, Pozos-Parra, Pilar, Sierra, Javier E.
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.09.2022
• Subjects: Deep learning; Forecasting; Machine learning; Review; Urban traffic flow; Deep learning; Urban traffic flow; Review; Forecasting; Machine learning
• ISSN: 2210-5379
• DOI: 10.1016/j.suscom.2022.100739

In recent decades, the development of transport infrastructure has had a great development, although traffic problems continue to spread due to increase due to the increase in the population in urban areas that require the use of these means of transport. This has led to increased problems related to congestion control, which has a direct impact on citizens: air pollution, fuel consumption, violation of traffic rules, noise pollution, accidents and loss of time. In Latin America, the disorderly growth of cities increases distances and routes, likewise, there is an accelerated increase in the number of cars and motorcycles, which increases the problem. In this sense, intelligent transport systems are an alternative to improve the traffic environment, they incorporate the internet of things and intelligent algorithms, for the collection of data from multiple sources and information processing, respectively, in order to improve the efficiency of the transport flow. However, the processing and modeling of traffic data is challenging due to the complexity of road networks, the space–time dependencies between them, and heterogeneous traffic patterns. In this review study, (i) the smart techniques used for the analysis of mobility data in the prediction of traffic flow in urban areas are grouped, likewise, (ii) the results of implementing said techniques are shown, in addition, (iii) The procedures performed are described and analyzed to understand the benefits and limitations of these smart techniques. Given the above, (iv) the data sets used in the literature and available for use are shown, in addition, (v) the quantifiable results of precision of the various techniques were compared, highlighting advantages and limitations, which allows us to (vi) identify the related challenges and, from there, (vii) propose a general taxonomy in which the knowledge acquired in this traffic flow review converges from a computational approach. •A bibliographic review of computational techniques for the prediction of urban traffic flow was carried out.•Various computational techniques used in the prediction of urban traffic flow are presented.•The precision results of the different urban traffic flow prediction techniques are compared.•Areas of opportunity are identified within the revised methods, which can be taken into account to help improve forecasting performance.