Predicting mobility choice and community connectivity in Latin America

Abstract

This study focuses on addressing the mobility challenges faced by developing regions of Latin America as data collection and the composition of formal and informal transportation. In this article, a tool is developed using a Machine Learning (ML) model that is able to learn and predict the patterns for choosing one mobility choice over another based on a student survey of the University of Guadalajara (UdeG). The study helps to understand which are the most relevant factors influencing mobility choice at one of the largest universities in Latin America, with travel time and number of household vehicles being the most determinant factors. The tool effectiveness is validated by the creation of two scenarios that simulate changes in mobility choices by relocating individuals closer to their destinations. The conducted experiment demonstrates a tendency towards walking and a significant decrease in private auto usage by relocating people closer to their destinations. The creation of this tool aims to help public institutions in making better decisions to develop a better society with reduced pollution, enhanced social impacts and climate change effects.