biomedical_data_generator.meta.compute_feature_strengths

biomedical_data_generator.meta.compute_feature_strengths(meta)[source]

Derive per-feature signal strengths from a DatasetMeta.

Returns three parallel length-n_features sequences: mean_strength (first-moment separation in standardized units), covariance_strength (range of per-class within-cluster correlation), and signal_channels (the active channels per feature).

A feature carries a signal iff at least one channel is active. Cluster columns are resolved one at a time via _cluster_column_strengths() (keyed on the column’s structural distance from its anchor), so the predicate on channels agrees with compute_feature_roles() per column — a feature is placed in an informative role iff its channels are non-empty, and in a noise role iff its channels are empty. A single cluster may therefore yield both kinds of column.

Parameters:: meta (DatasetMeta) – Resolved dataset metadata produced by biomedical_data_generator.generate_dataset().
Returns:: A FeatureStrengths instance with per-feature signal annotations.
Return type:: FeatureStrengths