In the same way that Big Data has transformed numerous industries by providing insights into many aspects, it has also been offering travelers and transportation authorities chances to understand transportation performance and issues better. Recently, the analysis of transportation data has attracted much attention, and many analytics methods have been proposed to extract insights from transportation data. The success of these methods is witnessed on capturing the performance on the major arterial roads because these roads are heavily equipped with sensors, traversed by many travelers, and the volume of information associated with these roads is high. However, in a city, most local roads do not have fixed sensors, and are not covered by enough floating sensors carried by travelers neither. The scarcity of sensory data brings great challenges in the accurate estimation of transportation information on these roads. As these roads occupy most areas in a city, it would be difficult to understand the city-wide transportation conditions with all these missing links.
To tackle the challenging problem of estimating transportation information for data-scarce roads, this project develops novel data mining methodologies that integrate heterogeneous urban data for the estimation of city-wide transportation information. Tasks include the inference of traffic speed, volume and emission, which are all critical components that contribute to the understanding of the overall transportation conditions in a city. The project also aims to detect traffic anomalies and derive a confidence measurement together with each estimate.
The successful completion of these tasks will: