Inference of parking search and location choice behavior using trip trajectories and Bayesian statistics

Abstract

Effectively designed parking solutions play a crucial role in mitigating urban challenges such as traffic congestion, excessive fuel consumption, and reduced safety. This study integrates GPS data with traveler survey data to model parking choice behavior, including preferences for parking type and choice for specific garage. Given the hierarchical nature of the choices, a nested logit model was estimated within a Bayesian framework using the Metropolis-Hastings algorithm. This approach is particularly efficient for high-dimensional models due to its fast convergence and straightforward implementation. The model was subsequently implemented within an agent-based transportation system simulation tool, POLARIS, to analyze the variations in drivers’ parking choice under diverse parking fee and capacity restriction scenarios. Analysis of the simulated scenarios for the Chicago network representing 8.5 million population revealed that while parking fee adjustments significantly impact travelers’ parking choices, fee variations during peak periods have minimal effect. For instance, a 50% rise in garage fees decreased the number of garage travelers by approximately 20%. However, raising garage fees by the same amount (50%) in the city during peak traffic periods led to a marginal drop of 7% in garage demand. Parking agencies generally set fees on the higher side to increase profit margins. However, parking policies should be designed to balance both travelers’ and transportation agencies’ interests in the long run. In addition, this study shows that allocating a portion of parking spaces to reserved parking or specific vehicle types, for a better curb space and parking management, significantly influences travelers’ decision-making in the parking choice selection.