Comparing Random Forests and Multinomial Logit Models for Urban Travel Mode Choice under Innovative Traffic Management Strategies

This study explores the factors affecting travel mode choices upon the implementation of two innovative Traffic Management Strategies in an urban setting, a Transit Signal Priority system and a Congestion Pricing scheme. To gather the necessary data, a stated preferences survey was conducted in three major European cities: Athens (Greece), Lisbon (Portugal), and Manchester (United Kingdom), aimed at the elicitation of individuals' transit mode choices. The collected data from each city were used to develop and calibrate multiple mode choice models, including econometric Discrete Choice (DC) models such as the Multinomial Logit (MNL), as well as state-of-the-art Machine Learning models like the Random Forest (RF) classifiers. Through the evaluation and comparison of the models’ results, an analysis of the factors influencing mode choice is presented, and the similarities and differences in the performance and interpretation of parametric and non-parametric models are discussed. The findings of this study can inform the development of sustainable transportation systems and contribute to more efficient decision-making processes in urban mobility management.