Beyond Aggregates: A Fine-Grained Analysis of Individual Mobility and Traffic Dependencies

Understanding mobile-user behavior requires joint modeling of mobility and traffic, as data consumption is shaped by where, when, and how users travel. Despite this clear intuition, most studies still treat the two in isolation, missing the intricate dependencies between them at the individual level. This paper propose a novel approach that explicitly captures the interplay between traffic and mobility behaviors using fine-grained mobile datasets. Using week-long eXtended Data Records (XDRs), we identify 13 interpretable features and pinpoint the mobility traits that truly drive traffic variation. These insights support a privacy-preserving user abstraction that represents each timeline as a sequence of discrete mobility–traffic states, capturing temporal dynamics and heterogeneity while generalizing across regions. We then introduce a probabilistic likelihood model that scores any mobility–traffic pairing, enabling cross-modality prediction and statistically sound fusion of fragmented logs. Experiments on four provincial datasets covering 1.3 million Chilean users show that the model reliably separates plausible from implausible behavior and generalizes from dense urban cores to mixed rural–urban contexts. The framework is descriptive, generative, and transferable, paving the way for anomaly detection, personalized QoE adaptation, and realistic network simulation.