| Description |
Recent years' advancements in sensing technology have generated an enormous amount of data in various fields and industries, including transportation. Public transportation systems, as a critical component within the transportation ecosystem, have also been experiencing much data growth. The availability of big data not only improves traditional transit service monitoring, but also enables high-resolution transit performance analysis that guides decision making. However, the potential of these datasets is not fully explored yet due to several challenges such as residing noises in data records and limited computational power. This dissertation tries to address three of those challenges: how to incorporate and analyze missing data due to lack of electronic footage, how to enable high-resolution performance measurements that require extensive computation, and how to interpret the high-resolution results? The first challenge was addressed in a quest to find missing data on the different fare payment methods without electronic footage, and their impact (among other factors) on bus Dwell Time (DT). Integrating information from multiple data sources, a combined approach of optimization and regression analysis was developed that offers a data-driven evaluation of existing fare payment structures and their individual effects on DT. Using the 35M bus rapid transit line operated by the Utah Transit Authority as a case study, the method demonstrates the robustness and strong predictive power in DT modeling. Then we introduce a new algorithm that is computationally elegant and mathematically efficient to address the second challenge of run-time reduction. An open-source toolbox written in C++ is developed to implement the algorithm. The toolbox is tested on the City of St. George's transit network to showcase dynamic transit accessibility analysis. The experimental evidence shows significant reduction on computational time. To address challenge three on interpreting the high-resolution transit accessibility results, the algorithm in the previous study was applied to the Salt Lake City's network to compute travel times at multiple departure times throughout the day. A series of indicators that are intuitive to interpret were developed to determine the varying causes of poor transit accessibility and identify areas with immediate needs for service improvements. This dissertation manifested that utilizing newly available datasets not only improves the resolution and accuracy of the transit service assessments, but also takes a step further to enable a comprehensive study of various factors (stop characteristics) impacting transit service efficiency and quantifying critical decision-making indices unveiling transit service effectiveness that were not possible before. Findings from this research are expected to lead to methodological advancements in data-driven approaches in public transit studies, and help transform the transit management mindset into a model of data-driven, sensing, and smart urban systems. |
