0.4 update #106
0.4 update #106
Conversation
run_unfold now checks the bf_vec input and runs either a mass univariate or an overlap corrected model. Note that there will be a discontinuity with Unfold.jl because the user has to provide a time window tuple = (start, stop) for mass univariate analysis.
This turns PyMNE into a weakdep and shifts `bids_load_data()` and `run_unfold` to an extension.
|
this looks great - but why is run_unfold dependent on PyMNE - only because of the extract_data? E.g. if I implement a custom file-loader and dont need preprocessing, shouldnt I be able to use run_unfold too? |
Yes, I thought about that. So far my only solution for this would be to check if the data in the DataFrame is a Matrix or a raw object (though we can't check for Python raw without PyMNE being loaded...), and if it's raw, then use extract data, otherwise use the Matrix as is |
|
You could have a catch-all that errors (with a hint to try loading pymne), a default case with providing matrices, and then the special case MNE Raw object? |
Updates the package to 0.4.
Breaking changes
run_unfold()changed. This was necessary to incorporate mass univariate modelingrun_unfoldis also part of the extensionNew feature
Addresses
#102
#101
TBD