sample.csv contains only synthetic records created for smoke tests and demonstrations.
Synthetic learner-state traces are included. A future adapter should map an event stream with learner, timestamp, action, and state labels into the sequence format used by this baseline.
See the header of sample.csv. Production adapters should validate types, missing values, timestamp semantics, and learner/session boundaries before analysis.
Personally identifiable information, raw student submissions, private LMS exports, proprietary course content, video/audio recordings, or licensed datasets that prohibit redistribution. Keep sensitive material outside Git and reference it through local paths or an approved secure store.
For any real experiment, record source, license/consent basis, population, collection period, exclusions, preprocessing, missingness, known biases, and permitted uses.