How to Pass additional argument to extract_features module in mne-feature

Support & Discussions
,
1

May I know what is the proper way to pass additional argument to the module extract_features.

Say for example, I would like to compute the power frequency, with the following parameters value.

normalize=False, ratios='all', psd_method='fft',

My understanding, this can be pass to the argument funcs_params.

So, I code something

        import numpy as np
        selected_features = ['pow_freq_bands']

        freq_bands = np.array ( [0.5, 4., 8.] )

        funcs_params = dict ( pow_freq_bands__normalize=False, pow_freq_bands__ratios='all',
                              pow_freq_bands__psd__method='fft', pow_freq_bands__freq_bands=freq_bands )
        features = extract_features ( epochs.get_data (), sfreq,
                                      selected_funcs=selected_features, return_as_df=True, funcs_params=funcs_params )

However, the compiler return an Valueerror

ValueError: Invalid parameter psd__method for transformer FeatureFunctionTransformer(freq_bands=array([0.5, 4. , 8. ]), normalize=True,
psd_method=‘welch’, psd_params=None, ratios=None). Check the list of available parameters using the get_params method of the transformer.

May I know how to properly pass the argument, if it is possible?

Im not sure whether this is the right platform to ask about mne-feature related question.

2

looks like you have double underscores in the wrong place:

pow_freq_bands__psd__method
                   ^^
3

Thanks for the eagle eye

4

The complete solution is

    import numpy as np
    selected_features = ['pow_freq_bands']

    freq_bands = np.array ( [0.5, 4., 8.] )

    funcs_params = dict ( pow_freq_bands__normalize=False,pow_freq_bands__ratios='all',pow_freq_bands__psd_method='fft',pow_freq_bands__freq_bands=freq_bands)
    features = extract_features ( epochs_zeroed.get_data (), sfreq,
                                  selected_funcs=selected_features, return_as_df=True, funcs_params=funcs_params )