Abstract
Mobile phone data provide valuable insights into urban structure, yet traditional clustering methods assign a single label per location, oversimplifying mixed-use areas. MAPLID introduces a supervised multi-label framework for place identification using Call Detail Records from Telecom Italia. Several multi-label classifiers are evaluated using stratified cross-validation. Label Powerset with Random Forest achieves 88.3% average precision on Milan, and validation in Trento confirms the framework’s ability to operate in more irregular and forested terrain. The results show that multi-label learning can capture concurrent urban functions and provide a computationally efficient basis for dynamic land-use mapping, urban planning and mobility analysis.
Citation
Mendoza-Hurtado, M., Romero-del-Castillo, J. A., & Ortiz-Boyer, D. (2026). MAPLID: a new multi-label approach for place identification using data supplied by mobile network operators. International Journal of Geographical Information Science. Online first. https://doi.org/10.1080/13658816.2026.2617932