Bicycle travel demand forecasting using geographic information systems and agent based modeling

With the increased recognition that bicycle transportation can be a viable mode choice, there is an expressed need for an accurate and easily assimilated bicycle transportation demand methodology. Moreover, the controversy surrounding bicycle forecasting is steeped with questions pertaining to accuracy, temporality, and repetition. Predominant bicycle travel demand methods include aggregate-level analysis, attitudinal surveys, discrete choice models, and regional travel models. Unfortunately, the aforementioned methods do not address the unpredictable and temporal nature of bicyclists, nor do they effectively account for the urban form-which has been documented to correlate to bicycling.;In this research a multi-methodological approach was employed to circumvent the apparent faults in discrete choice forecasting and determine if a bottom up methodology, acknowledging urban morphology, is valid. A binary logit model was tested for its effectiveness in bicycle mode choice selection, and then compared to a disaggregated approach using an Agent Based Model (ABM). To increase the authenticity of the ABM approach, Space Syntax model results and cognitive clues were integrated in the model to represent how urban morphology can influence bicyclist wayfinding. Space Syntax has the unique ability to describe computationally spatial patterns at the network level, which is not treated in discrete choice models, nor used in ABM previously. The outcomes of this research have exhibited the importance of urban morphology and cognition for bicyclists on two fronts. First, the binary logit model demonstrated that global and local visibility graph measures obtained from Space Syntax are strong predictors in the decision to bicycle to work. Several morphological factors were shown to be significant and increased overall model robustness when personal, household, environmental, and urban morphology factors were inserted. Secondly, the bicycle ABM developed here demonstrates that cognition and urban morphology play pivotal roles when bicycle agent patterns were analyzed against shortest path routes and estimated bicycle observations. The results of this research shed additional light on the ongoing investigations into the motivational factors involved in the decision to bicycle and how human cognition and urban design play a significant role in bicycle route selection. Moreover, the outcomes displayed in this research confirm the long standing notion that top down forecasting methods have much to be desired, and urban morphology are strong predictors of bicycle wayfinding.