Ica using picard

Mailing List Archive (read-only)
1

External Email - Use Caution

Hi,

I have updated mne to version 19. I am now playing around with ICA and I wanted to see how the Picard method would behave on my data. However, when I try to run this method using the following commands:

ica = ICA(n_components=picks.size, method=method)
# ica is fitted on epoched data (stored in self)
ica.fit(self, picks=picks, decim=decim)

I get a ModuleNotFoundError:

~/anaconda2/envs/mne3/lib/python3.7/site-packages/mne/preprocessing/ica.py in _fit(self, data, max_pca_components, fit_type)
    678 **self.fit_params)
    679 elif self.method == 'picard':
--> 680 from picard import picard
    681 _, W, _ = picard(data[:, sel].T, whiten=False,
    682 random_state=random_state, **self.fit_params)

ModuleNotFoundError: No module named 'picard'

What am I doing wrong here or is something wrong with my installation?

Cheers,

Dirk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20191003/8fa36b56/attachment.html

2

External Email - Use Caution

hi Dirk,

pip install python-picard

should fix it.

HTH
Alex

3

External Email - Use Caution

You should be able to `pip install python-picard` to get the `picard` ICA
package. If you used the `environment.yml` or `requirements.txt` from the
MNE repo to set up your env, it should have been included, though.

Eric