Análisis e Identificación de trayectorias de movilidad y modos de transporte utilizando datos de GPS

Abstract

Population growth has increased global traffic, intensifying greenhouse gas emissions. Therefore, the detection of transportation modes is crucial for identifying how people move in urban areas and designing strategies to reduce reliance on private vehicles. However, this process faces certain challenges, such as accurately differentiating between transportation modes with similar movement patterns, the multimodal nature of many trips, and the need for additional sensor data, which limits its applicability. To address these challenges, we have developed an automatic method for detecting transportation modes using GPS data through a supervised algorithm that analyzes travel characteristics. This document describes the methodology, from the development and implementation of the model to validation by mobility experts, including an analysis of mobility patterns through an interactive dashboard. A key finding is the automation of trip labeling using specific rules for each mode of transportation, which enables the construction of a machine learning model capable of determining the mode of transport. Additionally, the segmentation of trips allows the identification of walking segments to isolate single-mode trips and facilitate classification. Finally, the creation of detailed features from GPS points, such as speeds and accelerations, improves results even in simple models. For future research, it is recommended to add more features such as altitude, apply the process to more diverse datasets, and explore other supervised algorithms.