Current Status of Transportation Data Analytics and A Pilot Case Study Using Artificial Intelligence (AI)

Big data and data analytics recently have attracted substantial attention from both industry and academia. All state DOTs in the New England region own and generate a huge amount of data, including planning (e.g., transit ridership), operations (e.g., loop detector, traffic camera, SPaT), safety (e.g., crash records), and asset condition data (e.g., bridge and pavement conditions, traffic sign). These data sets are often collected and maintained by different divisions of DOTs, and such a practice may lead to duplication of efforts and underutilization of data. For example, sometimes one DOT division is not aware of the existence of a data set maintained by another division. This may result in the same data elements being collected and stored by multiple divisions. Also, data sets from different divisions could be but are usually not correlated for in-depth analysis. The power of the data is not fully understood and exploited as it should be. It is important for state DOTs to conduct a comprehensive review of their data, data needs, and data analysis and decision making practices, based on which the data collection, reduction, storage, and analysis efforts of different DOT divisions may be coordinated and unified, and be conducted in a more cost-effective manner. This integrated approach will help to turn data into useful information and produce valuable insights to support data-driven decision making. Such a process is also important for state DOTs in the New England region to learn from each other’s successful practices of transportation data analytics. While it is desirable to conduct a comprehensive review of all the data and data needs that state DOTs have, as the first step the focus of this problem statement is on data related to traffic operations, to ensure that the scope of work is feasible given the time and budget.