FR: Add "unit" parameter to get_data method for Epochs

UPDATE: https://github.com/mne-tools/mne-python/pull/9136 implements this for “raw” … we still need it for “epochs”

---

I often find myself doing things like this:

# get EEG data and scale to micro volts
# ntrials x nchannels x ntimes
eeg_data_uV = epochs.get_data() * 1e6

… just to do different kinds of analyses with custom functions.

I think we should add a "unit" parameter to .get_data():

"""
Parameters
----------
unit : str, one of ["nV, uV, mV, V"]
    The unit in which the data will be returned. Defaults to "V".
"""

This would make it more transparent for users what units the data are in they are receiving, and also offer a chance for customization.

I have a complete déjà vu writing up this issue, but I couldn’t find existing threads on this.

[larsoner]

I was wondering why the EOG channels are not considered as data channels in get_channel_types()

In some sense all channels are data channels. I think the idea is that the “data channels” in MNE correspond mostly to the channels that record brain data.

whether this units parameter should work on them?

I wouldn’t restrict it just to data channels. Any channels with units (which is most of them I think?) should be able to be modified by the unit parameter.

[sappelhoff]

The benefits are (at least) twofold:

1. when somebody (a non-user) reads your code, something like epochs.get_data() * 1e6 is much more confusing than epochs.get_data(unit=dict(eeg='uV')) --> where does the magic number 1e6 come from? They’d need to learn that in MNE all EEG data is represented in Volts, and that 1e6 scales to microVolts
2. as a fresh MNE user, I was confused about these units and a parameter unit would certainly have cleared up the situation for me

But I agree that this is an issue that requires quite some work, so no worries if you’d rather work on something else.