ezmsg.learn.process.slda

Shrinkage LDA classifier processor.

Note

This module supports the Array API standard via array_api_compat.get_namespace(). Input data is manipulated using Array API operations (permute_dims, reshape); a NumPy boundary is applied before sklearn.predict_proba.

Classes

class SLDA(*args, settings=None, **kwargs)

Bases: BaseTransformerUnit[SLDASettings, AxisArray, ClassifierMessage, SLDATransformer]

Parameters:

settings (Settings | None)

SETTINGS

alias of SLDASettings

class SLDASettings(settings_path, axis=None)

Bases: Settings

Parameters:

• settings_path (str)

• axis (str | None)

settings_path: str

axis: str | None = None

Deprecated since version 1.6.

Scheduled for removal in 2.0. The samples this classifies accumulate along one dimension, and the cached output template is keyed to it; that dimension now comes from stream_dim.

__init__(settings_path, axis=None)

Parameters:

• settings_path (str)

• axis (str | None)

Return type:

None

class SLDAState

Bases: object

axis: str = ''

The resolved stream dimension, fixed at reset so every later use agrees.

lda: LinearDiscriminantAnalysis

out_template: ClassifierMessage | None = None

class SLDATransformer(*args, **kwargs)

Bases: BaseStatefulTransformer[SLDASettings, AxisArray, ClassifierMessage, SLDAState]