Mathieu Scheltienne

Maybe, I didn't dig in too far, but it looked to me like the feature extraction code I translated was half-baked. And I think the unused `pct_data` variable appears both...

This is now too out of date, especially with the added `Spectrum` objects. Closing.

For real-time annotation, the causality of the filter prevents us from getting a clear relationship between a location on the viewing area and a location in the saved dataset. PR...

Depends on the overhaul of the Receiver API: - #60

I think that addition to the API would be very welcome. My initial impression is that it should mimic the `ClassArray` API from MNE-Python. Examples: - [RawArray](https://mne.tools/stable/generated/mne.io.RawArray.html) - [EpochsArray](https://mne.tools/stable/generated/mne.EpochsArray.html) -...

@RunKZhang I fixed my proposed class signature in the previous post, I defined it as a function for some reason :see_no_evil:

> Sorry I don't get why define it as a function. Since the predict method helps make segmentation of the raw object, and it is inherited from _BaseCluster, I think...

> But I think labels should be mandatory, since it is essential to calculate gev_ OK, then let's put it as third argument, after `info`. @vferat Sounds good to you?

OK, then let's go with this class signature: ``` class ClusterArray(_BaseCluster): def __init__( data: NDArray[float], info: Union[Info, ChInfo], cluster_names: Optional[List[str]] = None, fitted_data: Optional[NDArray[float]] = None, labels: Optional[NDArray[int]] = None,...

OK, that works for me.