filtering.filter_data#

autoclean.functions.preprocessing.filtering.filter_data(data, l_freq=None, h_freq=None, notch_freqs=None, notch_widths=0.5, method='fir', phase='zero', fir_window='hamming', verbose=None)[source]#

Filter EEG data using highpass, lowpass, and/or notch filtering.

This function applies digital filtering to continuous (Raw) or epoched EEG data. It supports highpass filtering to remove slow drifts, lowpass filtering to remove high-frequency noise, and notch filtering to remove line noise at specific frequencies (e.g., 50/60 Hz power line interference).

The function automatically detects the input data type and preserves it, returning the same type (Raw or Epochs) with identical structure but filtered time series data.

Parameters:

datamne.io.BaseRaw or mne.Epochs: The EEG data to filter. Can be any MNE Raw object (e.g., RawFIF, RawEEGLAB, etc.) or Epochs object.
l_freqfloat or None, default None: Low cutoff frequency for highpass filtering in Hz. If None, no highpass filtering is applied. Typical values: 0.1-1.0 Hz for removing slow drifts.
h_freqfloat or None, default None: High cutoff frequency for lowpass filtering in Hz. If None, no lowpass filtering is applied. Typical values: 30-100 Hz for removing high-frequency noise.
notch_freqslist of float or None, default None: Frequencies to notch filter in Hz. If None, no notch filtering is applied. Common values: [50] or [60] for power line noise, [50, 100, 150] for harmonics.
notch_widthsfloat or list of float, default 0.5: Width of the notch filter in Hz. If float, same width applied to all frequencies in notch_freqs. If list, must match length of notch_freqs. Wider notches remove more surrounding frequencies.
methodstr, default ‘fir’: Filtering method. ‘fir’ uses finite impulse response (linear phase), ‘iir’ uses infinite impulse response (faster but can introduce artifacts).
phasestr, default ‘zero’: Phase of the filter. ‘zero’ for zero-phase (no delay), ‘zero-double’ for zero-phase with double filtering, ‘minimum’ for minimum phase.
fir_windowstr, default ‘hamming’: Window function for FIR filter design. Options: ‘hamming’, ‘hann’, ‘blackman’. Affects filter characteristics and artifacts.
verbosebool or None, default None: Control verbosity of output. If None, uses MNE default.

Returns:

filtered_datamne.io.BaseRaw or mne.Epochs: The filtered data object, same type as input. Contains identical structure and metadata but with filtered time series data.