A GIS-Assisted Fuzzy Approach to Geographical Clustering of Mobile Phone Users’ Travel Behavior

Mobile phone usage data inherently involve many spatial elements; therefore, gathering extensive individual mobile phone records can offer unique insights into human spatial behavior at both personal and societal levels. This study contributes to travel behavior research by examining group-level human mobility obtained from millions of Hungarian mobile phone records. After developing mobility metrics from georeferenced cellular data, we applied a computationally efficient two- and three-dimensional Fuzzy C-Means (FCM) unsupervised clustering algorithm to identify groups of people with similar behavioral traits. The resulting membership probabilities—based on combinations of mobility metrics and user attributes—indicated that high travel distances or higher equipment prices could lead to a clear separation in travel behavior, while complex mobility patterns appeared less influenced by human factors such as age. Furthermore, even though the fuzzy outcomes offer probabilistic rather than exact group assignments, the generated maps revealed distinct, non-random spatial patterns.