Atlas, a subsystem of the Artemis learning management system used in large TUM computer science courses, implements competency-based education (CBE) and adaptive learning paths. Although the advantages of CBE are well documented, this reference implementation represents a novel, high-enrollment setting whose effects on students have not been examined. This thesis closes that gap with the first systematic evaluation of Atlas from the student perspective.
Using a mixed-methods design, we combine pre-/post-surveys with system log data to examine how students use and experience Atlas and to assess effects on exercise engagement and completion. Complementing prior instructor-centric work, we provide empirically grounded evidence on student behaviors and perceptions—evidence essential for informed design and responsible scaling of adaptive learning.
Findings indicate clear perceived benefits, usage clustering around key academic transitions, and associations with sustained course engagement and higher exam participation. Temporal scaffolding appears to structure self-directed learning more efficiently, aligning effort with instructional goals. The study offers a necessary empirical baseline for further evaluation and actionable guidance for deploying adaptive systems at scale.