Microscopic Analysis of Work Zone Delay using Big Data and Logistic Regression

Abstract
Work zone traffic management poses a significant challenge due to the need to minimize traffic delays while maintaining a safe environment for workers. In order to effectively manage traffic through work zones, it is necessary to understand how delays are affected by both internal and external factors, such as the work zone's design characteristics, road geometry, weather conditions, and traffic conditions. Despite the literature's attention to modeling work zone delays, most studies are limited to analyzing data from a small number of work zones and consider limited travel time observations. Macroscopic travel time prediction is also a key focus of existing literature. To address those gaps and provide a better understanding of factors affecting delay in work zones, this study uses a big dataset consisting of 15 million travel time observations collected every 2-3 minutes at 624 work zones located between the western borders of Alberta and Vancouver, BC. A microscopic delay analysis was performed whereby the impact of spatial, temporal, environmental, and segment-related variables on work zone delays was assessed using ordinal logistic regression. Based on the model statistics, high delays were associated with hourly traffic volume, peak hours, weekend travel, length of the work zone, and summertime, whereas precipitation affected delays in the opposite direction.