Data Driven Decision Making Tools for Transportation Work Zone Planning

This research provides tools and methods for integrating stakeholder input and crash data analytics to better guide transportation engineers in effective work zone design and management. Three key contributions are presented: the importance of stakeholder input in traffic management strategies, application of data mining and pattern recognition to identify high-risk drivers in work zones, and the use of multinomial logistic regression (MLR) as a tool to understand key findings from historic crash data. Work zone signage is mandated by the Manual on Uniform Traffic Control Devices (MUTCD), but the current configurations are often criticized by the driving public and state departments of transportation have questioned whether alternate signage would provide more cost-effective, equally safe options. A driving simulator study funded by the Missouri Department of Transportation (MoDOT) evaluated one such alternate sign configuration and determined that it received higher levels of driver satisfaction with no statistical impact on safety. Findings of driver preference for the alternate configuration are considered high value by MoDOT with respect to both mobility and safety. A second contribution focused on risk mitigation through data analytics. Pattern recognition and data mining techniques were applied to driving simulator data as part of a multi-criteria decision making tool to identify drivers with high risk potential. Findings related to age and gender suggest opportunities for driver education and training to increase safety. The third contribution identifies a method for analyzing historic crash data to determine key risk factors in fatality and serious injury accidents in work zones. Multinomial logistic regression (MLR) is used. Findings outline patterns and scenarios that should be integrated into work zone design to enhance safety and improve mobility with respect to work zone lighting, impact of weather, and the like.