<p>Learner profiling has gained significant attention in educational data mining due to the proliferation of online learning and big data analytics. However, most existing learner profiling approaches rely on static or coarse-grained representations, which fail to capture temporal dynamics and the interplay between individual profiles and group profiles. To address this limitation, this paper proposes a timeslot-based learner profiling method that models learners’ behavioral evolution at multiple temporal granularities while explicitly incorporating intra-timeslot group structures. This method constructs fine-grained timeslot profiles for individual learners and dynamically derives group profiles within each timeslot, enabling joint analysis of individual dynamics and collective learning patterns. Experiments conducted on a large-scale real-world online learning dataset demonstrate that the proposed method effectively captures meaningful behavioral evolution patterns and produces stable, interpretable learner group profiles. Furthermore, when applied to a downstream learning performance prediction task, the fused timeslot profiles consistently outperform static profiling baselines. The results indicate that explicitly modeling temporal evolution and group context provides more informative learner profiles, offering a robust foundation for dynamic learner analysis and data-driven educational decision support.</p>