Mobility situations in Mexico City Metropolitan Zone: An exploration of time and distance in the journey to work through machine learning

Abstract

This study examines the determinants of commuting time and distance using the mobility situations framework in the Mexico City Metropolitan Zone (MCMZ), a megacity marked by spatial mismatch, socioeconomic segregation, and fragmented transport infrastructure. Using data from the 2017 Origin-Destination Survey, we classify 192 travel districts into four mobility situations—Short Commutes, Long Commutes, Travelscarps, and Wormholes—based on average commuting time and distance. Our approach combines spatial econometrics with a machine learning LASSO algorithm to evaluate 88 potential predictors across transport infrastructure, urban spatial structure, and socioeconomic conditions. Results show that each situation is driven by distinct factors: Short Commutes align with centrality and privilege; Long Commutes with exclusion and mass transit; Travelscarps with inefficient short trips from poor infrastructure; and Wormholes with efficient long trips through multimodal strategies. The study demonstrates the value of the mobility situations framework in a Global South city and highlights machine learning’s utility for variable selection and theory-building in journey-to-work research.