This deliverable explores the state-of-the-art prediction methodologies such as data-driven and simulation-based estimation and prediction approaches, both on supply and demand side. The focus of the report is on deep learning approaches for efficient numerical representations of traffic networks, data availability and granularity, new data sources and graph theory. The analysis is widened with aspects such as transferability and generalizability of the different methods.
The state-of-the-art provides an overview of what has been done in the field of traffic state predictions and simulations and what should be further explored. The results of this exploration will provide the base for the development of the techniques for the remaining tasks of this WP.