This should fix the problem in #300, by returning the transposed version of the data from _OfflineStream._handle_data when the input is given in the form of a DataFrame. I also fixed the Error message that caused the confusion in the first place.
Additionally, this raises the issue of having 2 different expected data input layouts:
For numpy arrays: (n_channels, n_timepoints)
For pandas dataframes: (n_timepoints, n_channels)
This is AFAIK not very well explained in the documentation.
Changing the expected dimensions of the np.array is most likely impossible since it would break compatibility. But also changing the dimensionality of dataframes is probably also not a good idea since rows for datapoints and columns for features is the standard way people organize data.
This should fix the problem in #300, by returning the transposed version of the data from _OfflineStream._handle_data when the input is given in the form of a DataFrame. I also fixed the Error message that caused the confusion in the first place.
Additionally, this raises the issue of having 2 different expected data input layouts:
Changing the expected dimensions of the np.array is most likely impossible since it would break compatibility. But also changing the dimensionality of dataframes is probably also not a good idea since rows for datapoints and columns for features is the standard way people organize data.
So I have settled for this solution for now.