Deep Learning on Traffic Prediction Methods, Analysis and Future Directions

---

Abstract

Accurate traffic forecasting is essential for intelligent transportation frameworks. Course arranging, vehicle dispatch, and facilitating gridlock may all profit from precise traffic anticipating. This challenge is difficult due to the complex and dynamic spatiotemporal relationships that exist between various segments of the road network. This topic has recently received a lot of research attention, particularly the deep learning method, which has significantly improved traffic forecasting abilities. This work expects to introduce a far- reaching assessment of traffic prediction strategies in view of deep gaining according to different points of view. To get things started, we'll summarize and classify the current methods for traffic forecasting. Second, we discuss the most recent approaches utilized in various applications for traffic prediction. Thirdly, to help different specialists, we gather and organize generally utilized public datasets from the ongoing writing. In addition, we conduct extensive investigations to evaluate the display of various methodologies on a genuine public dataset in order to provide an evaluation and examination. Inconclusion, we examine irritating issues in this area.

---

Keywords

Traffic Prediction, Deep Learning, Spatial Temporal Dependency Modeling.