Key points are not available for this paper at this time.
Abstract Public transit systems in developing regions often face challenges due to rapid urbanization, limited resources, and a lack of comprehensive transit data, hindering effective strategic planning. This study proposes a framework for leveraging emerging data sources to guide transit network planning in data-limited environments. Using Greater-Maputo, Mozambique as a case study, we demonstrate how mobile phone location data, OpenStreetMap, and land use information can be utilized to extract key transit network components, including transit-viable road segments, high-demand stop locations, and efficient routes. We employ a modified semi-supervised self-training algorithm for transit viability prediction, density-based clustering for stop location extraction, and multi-objective metaheuristics for route extraction. The results show strong alignment with the operational GTFS data, capturing spatial patterns of transit suitability, identifying critical transit locations, redundancies, and underserved areas, and generating more direct and demand-aligned routes. The compiled extracted results in the form of GTFS data show a 17\% potential improvement in accessibility compared to the operational GTFS data. The proposed approach offers a potent and transferable methodology for data-driven transit planning, supporting the development of efficient, equitable, and sustainable transit systems in various contexts. This research contributes to the growing body of knowledge on evidence-based transit planning in data-scarce environments and lays the foundation for future research and policy interventions aimed at optimizing transit networks, enhancing accessibility, and promoting sustainable urban development in developing regions.
Mittal et al. (Fri,) studied this question.