biomedical_data_generator.meta.FeatureStrengths

class biomedical_data_generator.meta.FeatureStrengths(mean_strength, covariance_strength, signal_channels)[source]

Bases: object

Derived per-feature signal-strength annotation.

All sequences are length n_features and ordered by column index. Strengths are the generative (configured) effect sizes, not finite-sample estimates.

Parameters:

mean_strength (tuple[float, ...])
covariance_strength (tuple[float, ...])
signal_channels (tuple[tuple[str, ...], ...])

mean_strength

First-moment separation per feature in standardized units (see compute_feature_strengths for the unit caveat). 0.0 for any feature with no class-dependent mean signal.

Type:: tuple[float, …]

covariance_strength

Second-moment separation per feature, the range of the effective per-class within-cluster correlation. 0.0 for any non-cluster feature and any cluster with no class-dependent correlation.

Type:: tuple[float, …]

signal_channels

Per feature, the sorted active channels among ("covariance", "mean"); empty for noise features.

Type:: tuple[tuple[str, …], …]

__init__(mean_strength, covariance_strength, signal_channels)

Parameters:

mean_strength (tuple[float, ...])
covariance_strength (tuple[float, ...])
signal_channels (tuple[tuple[str, ...], ...])

Return type:

None

Methods

__init__(mean_strength, covariance_strength, ...)

Attributes

`mean_strength`
`covariance_strength`
`signal_channels`