Re-structuring the moabb core, implementing caching, creating tutorials, fixing some bugs and more #408
Conversation
TODO
|
|
Comments of Sylvain during meeting of June 30:
Do you have more comments @sylvchev ? |
There was a problem hiding this comment.
Hey @PierreGtch!
So fair is super okay, I reviewed 13 of 23 files, and I will need more time for the reviewing. I am directly changing the file to minimize your work.
Perfect, thanks for directly updating the files |
bruAristimunha
changed the title
|
Hey @PierreGtch! I REALLY liked how you defined the transformations for the basis of our paradigms, and I think it's cool how it was done! The new BID core structure is super good, I think the methodological design is clear, and all lines are clear in the context. I was not an easy PR to review haha I have some questions about the amount of testing, some documentation comments, why it might not be obvious in some cases. Congratulations on all the effort. I think this will make our reading and writing bottleneck a lot easier! |
|
Thank you @PierreGtch :) |
|
No problem, It was fun to do 😁 |
* add thielen2021 cvep dataset * add cvep paradigm * add thielen2021 cvep dataset tests * [pre-commit.ci] auto fixes from pre-commit.com hooks * update documentation related to addition c-VEP * Update cVEP paradigm according to the changes of #408 * change c-VEP paradigm to (potentially) multi-class * add c-VEP paradigm tests * add c-VEP banchmark tests * [pre-commit.ci] auto fixes from pre-commit.com hooks * update c-VEP documentation * remove redundant interval parameter from Thielen2021 dataset * remove redundant tmin/tmax tests from BaseCVEP * Add thielen2021 to datasets __init__.py * Add default interval * Complete scoring cases and add test * Reformat events in FakeDataset code * Fix coquille * Update fake cVEP datasets parameters * update documentation cvep * Add link in dataset_summary.rst * Refactoring the tests * Updating pytest * Fixing conflict with mne-bids and pytest * Increase the codacov * update documentation cvep * split functionality for reuse and update documentation * [pre-commit.ci] auto fixes from pre-commit.com hooks * add channel locations thielen2021 from loc file * Clarify paradigm doc (epoch level decoding) * Fixing conflict --------- Co-authored-by: Jordy Thielen <jordy.thielen@donders.ru.nl> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: PierreGtch <25532709+PierreGtch@users.noreply.github.com> Co-authored-by: Pierre Guetschel <pierre.guetschel@donders.ru.nl> Co-authored-by: bruAristimunha <a.bruno@aluno.ufabc.edu.br>
Closes #385
Closes #391 (BIDS conversion)
The goal of his PR is to save, on disk, a semi-processed version of the datasets to speed up their loading later.
The approach rn is to save the raw objects in a BIDS structure using mne_bids in EDF. We only apply the bandpass filters to the data before caching it. This way, the cached EDF files can then be read with
preload=Falseand only the epochs are loaded in memory.