Miles Matter: Demographics, Distance, and Decision-Making

Abstract

In this thesis, I investigate which variables have the strongest influence on an individual's travel mode choice depending on the purpose and level of urgency (leisure, essential, emergency) of the trip. I analyze the relationship between spatiotemporal costs conditioned by demographic segmentation using data on population mobility patterns in auto-centric Los Angeles and multimodal New York City. Through a synergistic three-pronged methodology consisting of spatial (time and distance analysis complemented by a spatial interaction model), statistical (multinomial logistic regression model), and machine learning-based (graph neural networks and extreme gradient boosting) analysis, I explore the multifaceted nature of decision-making processes in different urban environments. The hidden patterns revealed by artificial intelligence show that distance is the key determinant of mode choice, depending on the urban form of the city and its adaptation to multimodality.