eegproc.data.schema module

Column roles for signal samples and precomputed EEG features.

class eegproc.data.schema.EEGFrame(data, fs, kind='signal', subject_columns=('subject',), trial_columns=('trial',), time_column=None, label_columns=('label',), feature_columns=None)

Bases: object

An EEG table with explicit subject, trial, label, and feature columns.

Rows within a trial are sorted by time_column when supplied. If it is omitted, sample_index is used when present; otherwise row order is kept. Label and identifier columns are always excluded from inferred features.

Parameters:

• data (DataFrame)

• fs (float)

• kind (Literal['signal', 'features'], default: 'signal')

• subject_columns (tuple[str, ...], default: ('subject',))

• trial_columns (tuple[str, ...], default: ('trial',))

• time_column (str | None, default: None)

• label_columns (tuple[str, ...], default: ('label',))

• feature_columns (tuple[str, ...] | None, default: None)

data: DataFrame

feature_columns: tuple[str, ...] | None = None

fs: float

kind: Literal['signal', 'features'] = 'signal'

label_columns: tuple[str, ...] = ('label',)

subject_columns: tuple[str, ...] = ('subject',)

time_column: str | None = None

trial_columns: tuple[str, ...] = ('trial',)

class eegproc.data.schema.FeatureColumn(channel, band=None, feature=None)

Bases: object

Channel, optional band, and optional statistic in a feature name.

Parameters:

• channel (str)

• band (str | None, default: None)

• feature (str | None, default: None)

band: str | None = None

channel: str

feature: str | None = None

eegproc.data.schema.median_split(label_column, *, threshold)

Create a fixed-threshold label transform; no threshold is fitted to data.

Parameters:

• label_column (str)

• threshold (float)

eegproc.data.schema.parse_feature_column(name)

Parse channel[_band][_statistic] without substring matching.

Parameters:

name (str)

Return type:

FeatureColumn