Planning for bicycle parking: Predicting demand using stated preference and count data

Abstract

Predicting bicycle parking demand is critical to optimizing parking facilities and thereby promoting cycling. Unfortunately, previous studies have not considered facility type and location when predicting bicycle parking demand, which is critical to meeting user needs, especially in scenarios with multiple parking options, such as on university campuses, as in our case study. The paper presents a predictive model for bicycle parking demand using a synthetic population derived from building space utilization data, a mobility survey, parking facility data, and results from a stated preference experiment on bicycle parking preferences. We evaluate the model’s quality using count data from 2022 and 2023 and the influence of including facility types (front wheel racks, u-racks, bicycle parking stations) and whether they are covered. We also analyze the influence of beeline-based distances to reach a facility and to get from the facility to the destination and examine how to weigh them.

Incorporating facility types and coverage substantially improves the model’s predictive accuracy, but only if the model’s sensitivity to walking distances between facilities and buildings is increased. This suggests that stated preference experiments on bicycle parking choice behavior may underestimate cyclists’ sensitivity to walking distances. In contrast, accounting for cycling detours to reach a facility does not contribute to prediction quality. Thus, when cyclists have multiple parking options, it is crucial to consider walking distances for realistic predictions. Furthermore, user-centered planning requires careful consideration of parking facility attributes and the specific preferences of target cyclist groups when determining the size and location of parking facilities.