Statistical Methods for the Evaluation and Monitoring of Traveler Information System Data Quality

The size and complexity of modern transportation networks have created new challenges for the planning and operation of transportation infrastructure. Advancements in sensor technologies have greatly assisted civil engineers with this task. However, along with advances in sensor technologies come new challenges in understanding the application and scope of the data generated by these technologies. One of the most promising sensing technologies today can generate estimates of link travel time from probe-vehicle samples. These estimates of link travel time can be used for a number of important applications in transportation engineering such as performance measurement. However, these estimates of travel time rely on sample observations of traffic parameters from a variety of sources over large spatial and temporal extents. Therefore, data quality validation and monitoring of travel time estimates is an important task that civil engineers will require today and in the future. This dissertation presents an overview of the steps in a data quality evaluation and proposes new methods in three critical areas: link benchmark estimation, link selection, and data quality monitoring. The methods build on existing statistical techniques and apply these methods to the problem of Traveler Information System data quality evaluation and monitoring.