We have created a representative dataset based on the collection of different traffic metrics from VANET simulations in urban scenarios. The five collected metrics that compose the dataset used to train and test our ML-based forwarding algorithm are available bandwidth, distance to destination, vehicles’ density, MAC layer losses, and vehicle’s trajectory. Notice that those metrics are gathered by the vehicles from the beacons periodically interchanged with the vehicles in their neighborhood (i.e., with vehicles within their transmission range). This way, nodes have local knowledge of the VANET, according to the decentralized nature inherent in VANETs.