Predicting the Fluctuation of Travel Time Reliability as a Result of Congestion Variations by Bagging-Based Regressors

Travel time reliability affects the behavior of passengers in private or public transportation and can be seen as an important factor in the context of freight transportation. The main cause of travel time oscillation, known as travel time reliability, is congestion. Congestion is classified into two categories: recurring and nonrecurring. Recurring congestion, which is the topic of this study, is formed when supply surpasses capacity. Peak periods are good examples of recurring congestion. In this paper, by utilizing different bagging regressor methods, the effect of speed flow reduction, compared to Free Flow Speed (FFS) in terms of congestion was studied on the Planning Time Index (PTI) on a section of Interstate 64 in the United States (US). Then, by analyzing PTI changes based on congestion variation, it was revealed that when speed reduction surpasses 10%, travel time leaves its reliability. Also, when the congestion is somewhere around 0.7 to 0.75, the unreliability becomes severe. These findings were directly extracted from scatter plots drawn by bagging and bootstrapping samples which were used to improve the accuracy of PTI prediction.