pyeeg.features.reduction.ReductionConfig

class pyeeg.features.reduction.ReductionConfig(method: str = 'pca', n_components: int | None = None, variance_threshold: float = 0.95, random_state: int | None = None, whiten: bool = False)

Configuration for feature reduction.

method

Reduction method to use. One of "pca" (principal component analysis), "ica" (independent component analysis via whitening), or "none" (identity transformation). Defaults to "pca".

Type:

str

n_components

Number of components to retain. If None, the number is derived from variance_threshold (PCA) or kept equal to the input dimensionality (ICA). Defaults to None.

Type:

Optional[int], optional

variance_threshold

For PCA with n_components=None: the cumulative explained-variance ratio at which the component count is selected. Defaults to 0.95.

Type:

float

random_state

Seed for reproducible results. Currently reserved: the implemented PCA/ICA routines are deterministic and do not use this field. Defaults to None.

Type:

Optional[int], optional

whiten

Whether to whiten the reduced components. Currently reserved: the implemented routines do not apply an explicit whitening step (ICA internally whitens as part of its construction). Defaults to False.

Type:

bool

Methods