Paper detail

Improving Data Quality in Intelligent Transportation Systems

Intelligent Transportation Systems (ITS) use data and information technology to improve the operation of our transportation network. ITS contributes to sustainable development by using technology to make the transportation system more efficient; improving our environment by reducing emissions, reducing the need for new construction and improving our daily lives through reduced congestion. A key component of ITS is traveler information. The Oregon Department of Transportation (ODOT) recently implemented a new traveler information system on selected freeways to provide drivers with travel time estimates that allow them to make more informed decisions about routing to their destinations. The ODOT project aims to improve traffic flow and promote efficient traffic movement, which can reduce emissions rates and improve air quality. The new ODOT system is based on travel data collected from a recently-increased set of sensors installed on its freeways. Our current project investigates novel data cleaning methodologies and the integration of those methodologies into the prediction of travel times. We use machine learning techniques on our archive to identify suspect data, and calculate revised travel times excluding this suspect data. We compare the resulting travel time predictions to ground-truth data, and to predictions based on simple, rule-based data cleaning. We report on the results of our study using qualitative and quantitative methods.