[Mne_analysis] question about epochs.next()

Wed Apr 6 01:42:17 EDT 2016

>
> I see. Sounds like that would work, I'll give it a try. thanks.
>
> FWIW, from the POV of a new mne-python user, the examples quite
> prominently use "evoked" to mean "average of epochs" (e.g.,
> http://martinos.org/mne/dev/auto_tutorials/plot_introduction.html?highlight=evoked).
> I probably missed the Epochs.iter_evoked method() on the basis of its name
> (because I didn't want averages, and it wasn't immediately obvious that
> "nave=1" meant "n_epochs_to_average=1", i.e., no averaging).
>
well point taken. The rationale for this comes from the realtime code where
epochs arrive as a stream and we want to add them with "+" to compute an
online average. This '+' operator is only offered by the Evoked class.

if the iter_* function returns Epochs with a single epoch then you need to
call average to get an Evoked. In this case you need to make 2 copies of
the measurement info to avoid weird bugs due to inplace changes. It was
also my performance driven decision to have iter return the data and not
Epochs as it requires no new copy of the measurement info.

What this discussion tells me is that the iterator on the data should be
more explicit eg.

epochs.iter_data()

let's continue this discussion at :
https://github.com/mne-tools/mne-python/issues/3099

Alex

> So now I'm doubly-puzzled: why does Epochs.next() not return a single
> epoch, and why does Epochs.iter_evoked() not return an average of epochs?
> I understand the need to preserve some backwards-compatibility, but the
> naming conventions here seem very ad-hoc to me and worth changing prior to
> a 1.0 release.  Until then, I guess I'll just change the docstrings to
> match what these methods actually do.
> On Apr 5, 2016 5:11 PM, "Mainak Jas" <mainakjas at gmail.com> wrote:
>
>> Sorry if I wasn't clear but iter_evoked() gives you a single epoch -- not
>> an average. It has nave=1.
>>
>> Mainak
>>
>> On Wed, Apr 6, 2016 at 1:45 AM, Dan McCloy <drmccloy at uw.edu> wrote:
>>
>>> thanks Mainak, but my use case is a little strange, and I don't think
>>> iter_evoked will help.  I need potentially different crop windows for each
>>> event type (same length, but different start/end points), but I need to
>>> keep the individual epochs (not averages) because they're getting fed into
>>> a classifier.  I'm happy to do a PR to change the docs if the consensus is
>>> that it is indeed the docs that should change, not the implementation of
>>> next().
>>>
>>> On Tue, Apr 5, 2016 at 4:27 PM, Mainak Jas <mainakjas at gmail.com> wrote:
>>>
>>>> Hi Dan,
>>>>
>>>> Since you are iterating over single epochs, it might make more sense to
>>>> use the iterator which returns an evoked object. You can do something like:
>>>>
>>>> for ev in epochs.iter_evoked()
>>>>     ev.crop(tmin, tmax)
>>>>
>>>> Does this help? Fix to the docs is more than welcome.
>>>>
>>>> Best regards,
>>>> Mainak
>>>>
>>>>
>>>> On Tue, Apr 5, 2016 at 10:05 PM, Dan McCloy <drmccloy at uw.edu> wrote:
>>>>
>>>>> On Tue, Apr 5, 2016 at 12:19 PM, Chris Holdgraf <
>>>>> choldgraf at berkeley.edu> wrote:
>>>>>
>>>>>> sometimes it's more useful to return arrays.
>>>>>>
>>>>>
>>>>> Sure, but that's why there is a get_data method, right?  IMO the
>>>>> docstring reflects how I expected it to work before even reading the
>>>>> docstring, so my vote would be to make .next() actually do that, rather
>>>>> than changing the docstring to reflect what it does currently.  Issue
>>>>> created: https://github.com/mne-tools/mne-python/issues/3099
>>>>>
>>>>>
>>>>>> On Tue, Apr 5, 2016 at 12:03 PM, Dan McCloy <drmccloy at uw.edu> wrote:
>>>>>>
>>>>>>> I was puzzled today why this works...
>>>>>>>
>>>>>>> for ix, _ in enumerate(epochs):
>>>>>>>     new_epoch = epochs[ix].crop(tmin, tmax, copy=True)
>>>>>>>
>>>>>>> ... but this does not:
>>>>>>>
>>>>>>> for epoch in epochs:
>>>>>>>     new_epoch = epoch.crop(..., copy=True)
>>>>>>>
>>>>>>> AttributeError: 'numpy.ndarray' object has no attribute 'crop'
>>>>>>>
>>>>>>> According to the docs, Epochs.next() is supposed to return an
>>>>>>> instance of the epochs object, but it doesn't: looking at the code, it
>>>>>>> actually returns something like self._data[self._current].  I would have
>>>>>>> expected it to return something like self.__get_item__(self._current).  So
>>>>>>> are the docs wrong, or is the definition of Epochs.next() implemented wrong?
>>>>>>>
>>>>>>> Daniel McCloy
>>>>>>> http://dan.mccloy.info/
>>>>>>> Postdoctoral Research Associate
>>>>>>> Institute for Learning and Brain Sciences
>>>>>>> University of Washington
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Mne_analysis mailing list
>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>
>>>>>>>
>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>> whom it is
>>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>>> the e-mail
>>>>>>> contains patient information, please contact the Partners Compliance
>>>>>>> HelpLine at
>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>>>> you in error
>>>>>>> but does not contain patient information, please contact the sender
>>>>>>> and properly
>>>>>>> dispose of the e-mail.
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _____________________________________
>>>>>>
>>>>>> PhD Candidate in Neuroscience | UC Berkeley <http://hwni.org/>
>>>>>> Editor and Web Director | Berkeley Science Review
>>>>>> <http://sciencereview.berkeley.edu/>
>>>>>> _____________________________________
>>>>>>
>>>>>> _______________________________________________
>>>>>> Mne_analysis mailing list
>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>
>>>>>>
>>>>>> The information in this e-mail is intended only for the person to
>>>>>> whom it is
>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>> the e-mail
>>>>>> contains patient information, please contact the Partners Compliance
>>>>>> HelpLine at
>>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>>> you in error
>>>>>> but does not contain patient information, please contact the sender
>>>>>> and properly
>>>>>> dispose of the e-mail.
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Mne_analysis mailing list
>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>
>>>>>
>>>>> The information in this e-mail is intended only for the person to whom
>>>>> it is
>>>>> addressed. If you believe this e-mail was sent to you in error and the
>>>>> e-mail
>>>>> contains patient information, please contact the Partners Compliance
>>>>> HelpLine at
>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>> you in error
>>>>> but does not contain patient information, please contact the sender
>>>>> and properly
>>>>> dispose of the e-mail.
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> Mne_analysis mailing list
>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>
>>>>
>>>> The information in this e-mail is intended only for the person to whom
>>>> it is
>>>> addressed. If you believe this e-mail was sent to you in error and the
>>>> e-mail
>>>> contains patient information, please contact the Partners Compliance
>>>> HelpLine at
>>>> http://www.partners.org/complianceline . If the e-mail was sent to you
>>>> in error
>>>> but does not contain patient information, please contact the sender and
>>>> properly
>>>> dispose of the e-mail.
>>>>
>>>>
>>>
>>> _______________________________________________
>>> Mne_analysis mailing list
>>> Mne_analysis at nmr.mgh.harvard.edu
>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>
>>>
>>> The information in this e-mail is intended only for the person to whom
>>> it is
>>> addressed. If you believe this e-mail was sent to you in error and the
>>> e-mail
>>> contains patient information, please contact the Partners Compliance
>>> HelpLine at
>>> http://www.partners.org/complianceline . If the e-mail was sent to you
>>> in error
>>> but does not contain patient information, please contact the sender and
>>> properly
>>> dispose of the e-mail.
>>>
>>>
>>
>> _______________________________________________
>> Mne_analysis mailing list
>> Mne_analysis at nmr.mgh.harvard.edu
>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>
>>
>> The information in this e-mail is intended only for the person to whom it
>> is
>> addressed. If you believe this e-mail was sent to you in error and the
>> e-mail
>> contains patient information, please contact the Partners Compliance
>> HelpLine at
>> http://www.partners.org/complianceline . If the e-mail was sent to you
>> in error
>> but does not contain patient information, please contact the sender and
>> properly
>> dispose of the e-mail.
>>
>>
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>
>
> The information in this e-mail is intended only for the person to whom it
> is
> addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> contains patient information, please contact the Partners Compliance
> HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in
> error
> but does not contain patient information, please contact the sender and
> properly
> dispose of the e-mail.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20160406/ef66d9cd/attachment.html 