A comprehensive review of traffic flow prediction: from traditional models to deep learning architectures

Traffic congestion gradually becomes an increasingly serious problem in cities. Intelligent transportation systems (ITS) have been introduced to alleviate traffic congestion. Traffic flow prediction, as an important part of ITS, has become a research hotspot in recent years. This study presents machine learning techniques in traffic flow prediction. First, it systematically introduces the task of traffic flow prediction, including the research background and problem definition. Then, it reviews twelve benchmark datasets commonly used for traffic flow prediction and provides a detailed description of the data processing steps. After that, this study highlights the theoretical foundations and significant impact of Graph Convolutional Networks for traffic flow modeling. Finally, this study provides an in-depth analysis of the opportunities and challenges in this field and offers pertinent suggestions for future research. Overall, this study can help researchers quickly get started with traffic flow prediction.