ENH: Add models for ds003, ds114 #160
Conversation
* Initial capitals in JSON keys * snake_case for variables * AutoContrasts for dataset-level analysis * High-pass filter cutoff
* Initial capitals in JSON keys * snake_case for variables * AutoContrasts for dataset-level analysis * High-pass filter cutoff Additionally, moved Split transformation to subject level, eliminating session-level step
|
Thanks, you seem to have made this PR from your old fork ... are you sure that this will not reintroduce some bad history? Why not delete your old fork, make a fresh one, and proceed? |
|
I rebased on the current master. |
|
I see ... I also just realized that we won't introduce bad history via PRs, but rather via direct pushes to Is this ready to be merged? If you need a review, you'd have to tag a member that has some experience with the models spec (I am not familiar enough with that). |
|
No, it's possible for someone to introduce bad history, but it should show up as a ton of commits in the PR. Typically just asking them to rebase on I'll request review from @tyarkoni and @adelavega. |
|
I looked through the models (admittedly not very carefully, but short of trying to run them through pybids and fitlins, I don't know that my careful review is terribly helpful) and they LGTM. Edit: one caveat is that the first model is missing intermediate steps between run and dataset, which I believe means that the dataset level should expect to receive multiple runs per session and/or subject. My recollection is that currently aggregating within these intermediate levels must be explicit (e.g., via |
There was a problem hiding this comment.
to get this PR merged, a .SKIP_VALIDATION file (like in the derivatives ds) should be added to the ds003 and ds114 datasets, ... or the models directories added in this PR should be .bidsignored
|
Should live in bids stats model zoo |
Replaces #130.