Repository navigation
Keep client study data out of the public repo and add a data manifest #227
Description
Is your feature request related to a problem? Please describe.
Dr. Gajapriya sent 19 telemetry files from a human-participant study. The repo is public, and inference/Data/ has held committed CSVs before. Nothing stops someone from committing these files by accident.
Describe the solution you'd like
- Add
inference/Data/client/to.gitignore, along with every generated CSV (*_features.csv,pilot_clusters.csv,group_centres.csv). The second half was also raised in the 215 deviation features #218 review. - Commit
inference/Data/client/MANIFEST.md. It lists file names, the pilot/trial index, row counts, a SHA-256 per file and where the team can get the files (a shared drive), and contains no data. - Ask Dr. Gajapriya whether derived per-flight features can be committed, and whether there is an IRB or data-use condition we need to follow.
Acceptance criteria
-
git statusstays clean after copying the files in and running the whole pipeline - The manifest lists all 19 files with their hashes
- The client's answer on sharing derived features is recorded in this issue
Activity
jmacd867 commentedon Oct 15, 2026
One note before this lands:
.gitignorehalf doesn't depend on the client's answer and can land now. The data-handling questions going to the client at the upcoming meeting: is there an IRB protocol or data-use agreement, what does it require (storage, access, deletion), are students covered by it or do we need to be added and complete human-subjects training, and may derived data (per-flight features, group assignments) be committed.jmacd867 commentedon Oct 16, 2026
Answers from the client meeting (`docs/client_meeting_notes.md`):
Still not answered: whether derived per-flight features can be committed, whether the students must be added to the protocol or complete human-subjects training, and storage, access and deletion rules. The `.gitignore` half can land now.