Skip to content

[MRG] Require good and bad channels when creating a SpectrumArray object#12877

Merged
mscheltienne merged 7 commits intomne-tools:mainfrom
mscheltienne:fix-spectrum
Oct 1, 2024
Merged

[MRG] Require good and bad channels when creating a SpectrumArray object#12877
mscheltienne merged 7 commits intomne-tools:mainfrom
mscheltienne:fix-spectrum

Conversation

@mscheltienne
Copy link
Copy Markdown
Member

@mscheltienne mscheltienne commented Sep 30, 2024 •
edited
Loading

Failure reported today:

import numpy as np
from mne.datasets import sample
from mne.io import read_raw
from mne.time_frequency import SpectrumArray
from numpy.testing import assert_allclose

fname = sample.data_path() / "MEG" / "sample" / "sample_audvis_raw.fif"
raw = read_raw(fname, preload=False).pick("eeg").load_data()

spectrum1 = raw.copy().crop(0, 10).compute_psd()
spectrum2 = raw.copy().crop(10, 20).compute_psd()

# make sure you have the same freqs!
assert_allclose(spectrum1.freqs, spectrum2.freqs)

# array of shape (n_channels, n_freqs)
data1 = spectrum1.get_data()
data2 = spectrum2.get_data()
data = np.array([data1, data2])  # array of shape (n_spectrum, n_channels, n_freqs)
data = np.average(data, axis=0)

# recompute the Spectrum object
spectrum = SpectrumArray(data, spectrum1.info, spectrum1.freqs)
spectrum.plot()

Raises IndexError: index 59 is out of bounds for axis 0 with size 59 in spectrum.plot, i.e. the plotting functions expects the good and bad channels. And indeed, spectrum1 or spectrum2 include the bads, i.e. the underlying array as shape (60, ...).

Now, if instead you include the bad channels:

data1 = spectrum1.get_data(exclude=())
data2 = spectrum2.get_data(exclude=())

It will also raise but this time in the creation of the SpectrumArray: ValueError: The number of channels in data(60) must match the number of good data channels ininfo (59).

There is a clear inconsistency in what SpectrumArray expects and BaseSpectrum.plot expects. I think the cleaner fix is to have SpectrumArray require the good and bad channels; matching the behavior obtained by Spectrum when running inst.compute_psd(...).

@larsoner
Copy link
Copy Markdown
Member

larsoner commented Sep 30, 2024

This one looks real:

FAILED mne/time_frequency/tests/test_spectrum.py::test_spectrum_array_errors - AssertionError: Regex pattern did not match.
 Regex: 'number of channels.*good data channels'
 Input: 'The number of channels in `data` (4) must match the number of good + bads data channels in `info` (5).'

@mscheltienne mscheltienne changed the title Require good and bad channels when creating a SpectrumArray object [MRG] Require good and bad channels when creating a SpectrumArray object Oct 1, 2024
drammock
drammock approved these changes Oct 1, 2024
View reviewed changes
Copy link
Copy Markdown
Member

@drammock drammock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

just a few typos. LGTM!

mscheltienne and others added 2 commits October 1, 2024 17:13
@mscheltienne @drammock
apply review [skip ci]
bbf86ec 
Co-authored-by: Daniel McCloy <dan@mccloy.info>
@mscheltienne
fix changelog
833b39b
drammock
drammock reviewed Oct 1, 2024
View reviewed changes
mscheltienne and others added 2 commits October 1, 2024 17:22
@mscheltienne @drammock
update changelog
5107c26 
Co-authored-by: Daniel McCloy <dan@mccloy.info>
@mscheltienne
fix tests
112b833
@mscheltienne mscheltienne enabled auto-merge (squash) October 1, 2024 16:37
@mscheltienne mscheltienne merged commit 7af60e6 into mne-tools:main Oct 1, 2024
@mscheltienne mscheltienne deleted the fix-spectrum branch October 2, 2024 07:40
larsoner added a commit to larsoner/mne-python that referenced this pull request Oct 28, 2024
@larsoner
Merge remote-tracking branch 'upstream/main' into feature/TD-nirs_snirf
34c36f5 
* upstream/main: (824 commits)
  Add `psd_args` to `plot_ica_sources` and `ICA.plot_sources` (mne-tools#12912)
  Fix GDF NumPy >= 2 (mne-tools#12909)
  [pre-commit.ci] pre-commit autoupdate (mne-tools#12908)
  ENH: Improve report usability (mne-tools#12901)
  MAINT: Avoid problematic PySide6 (mne-tools#12902)
  Sync README dependencies with pyproject.toml (mne-tools#12890)
  remove trailing slash from pybv base URL [ci skip] (mne-tools#12892)
  Cast tuple of filenames to list to improve error handling (mne-tools#12891)
  Website (mne-tools#12885)
  [pre-commit.ci] pre-commit autoupdate (mne-tools#12888)
  BUG: Fix bugs with coreg (mne-tools#12884)
  Bump mamba-org/setup-micromamba from 1 to 2 in the actions group (mne-tools#12887)
  Update spacing for comments in pyproject.toml (mne-tools#12886)
  make HTML repr for Forward match others (mne-tools#12883)
  MAINT: Linkchecks [circle deploy] (mne-tools#12882)
  Update roadmap (mne-tools#12872)
  [MRG] Require good and bad channels when creating a SpectrumArray object (mne-tools#12877)
  [pre-commit.ci] pre-commit autoupdate (mne-tools#12879)
  MAINT: Update code credit (mne-tools#12880)
  BUG: Fix bug with Path casting (mne-tools#12878)
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@drammock drammock drammock approved these changes

@larsoner larsoner Awaiting requested review from larsoner larsoner is a code owner

@agramfort agramfort Awaiting requested review from agramfort agramfort is a code owner

@dengemann dengemann Awaiting requested review from dengemann

@adam2392 adam2392 Awaiting requested review from adam2392

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@mscheltienne @larsoner @drammock