Abstract

Traffic management is being more important than ever, especially in overcrowded big cities with over-pollution problems and with new unprecedented mobility changes. In this scenario, road-traffic prediction plays a key role within Intelligent Transportation Systems, allowing traffic managers to be able to anticipate and take the proper decisions. This paper aims to analyze the situation in a commercial real-time prediction system with its current problems and limitations. We analyze issues related to the use of spatiotemporal information to reconstruct the traffic state. The analysis unveils the trade-off between simple parsimonious models and more complex models. Finally, we propose an enriched machine learning framework, Adarules, for the traffic state prediction in real-time facing the problem as continuously incoming data streams with all the commonly occurring problems in such volatile scenario, namely changes in the network infrastructure and demand, new detection stations or failure ones, among others. The framework is also able to infer automatically the most relevant features to our end-task, including the relationships within the road network, which we call as “structure learning”. Although the intention with the proposed framework is to evolve and grow with new incoming big data, however there is no limitation in starting to use it without any prior knowledge as it can starts learning the structure and parameters automatically from data. (Part of special issue: 20th EURO Working Group on Transportation Meeting, EWGT 2017, 4-6 September 2017, Budapest, Hungary) Peer Reviewed

Abstract

The uncertainty of the take-off time is a major contribution to the loss of trajectory predictability. At present, the Estimated Take-Off Time (ETOT) for each individual flight is extracted from the Enhanced Traffic Flow Management System (ETFMS) messages, which are sent each time there is an event triggering a recalculation of the flight data by the Network Man- ager Operations Centre. However, aircraft do not always take- off at the ETOTs reported by the ETFMS due to several factors, including congestion and bad weather conditions at the departure airport, reactionary delays and air traffic flow management slot improvements. This paper presents two machine learning models that take into account several of these factors to improve the take- off time prediction of individual flights one hour before their estimated off-block time. Predictions performed by the model trained on three years of historical flight and weather data show a reduction on the take-off time prediction error of about 30% as compared to the ETOTs reported by the ETFMS. Peer Reviewed