biomedical_data_generator.DatasetMeta

class biomedical_data_generator.DatasetMeta(feature_names, informative_idx, noise_idx, corr_cluster_indices, anchor_idx, standalone_informative_groups, standalone_noise_range, mean_per_class_effect, covariance_per_class_correlation, baseline_correlation, cluster_label, cluster_structure, cluster_proxy_attenuation, n_classes, class_names, samples_per_class, class_sep, batch=None, random_state=None, resolved_config=<factory>)

Bases: object

Metadata about the generated dataset.

This captures the resolved ground-truth structure of the dataset (feature roles, cluster layout, anchor properties) plus a snapshot of the generator configuration.

Parameters:

• feature_names (list[str])

• informative_idx (list[int])

• noise_idx (list[int])

• corr_cluster_indices (dict[int, list[int]])

• anchor_idx (dict[int, int])

• standalone_informative_groups (tuple[StandaloneGroupMeta, ...])

• standalone_noise_range (tuple[int, int])

• mean_per_class_effect (dict[int, dict[int, float] | None])

• covariance_per_class_correlation (dict[int, dict[int, float] | None])

• baseline_correlation (dict[int, float])

• cluster_label (dict[int, str | None])

• cluster_structure (dict[int, Literal['equicorrelated', 'toeplitz']])

• cluster_proxy_attenuation (dict[int, float])

• n_classes (int)

• class_names (list[str])

• samples_per_class (dict[int, int])

• class_sep (list[float])

• batch (BatchMeta | None)

• random_state (int | None)

• resolved_config (dict[str, object])

__init__(feature_names, informative_idx, noise_idx, corr_cluster_indices, anchor_idx, standalone_informative_groups, standalone_noise_range, mean_per_class_effect, covariance_per_class_correlation, baseline_correlation, cluster_label, cluster_structure, cluster_proxy_attenuation, n_classes, class_names, samples_per_class, class_sep, batch=None, random_state=None, resolved_config=<factory>)

Parameters:

• feature_names (list[str])

• informative_idx (list[int])

• noise_idx (list[int])

• corr_cluster_indices (dict[int, list[int]])

• anchor_idx (dict[int, int])

• standalone_informative_groups (tuple[StandaloneGroupMeta, ...])

• standalone_noise_range (tuple[int, int])

• mean_per_class_effect (dict[int, dict[int, float] | None])

• covariance_per_class_correlation (dict[int, dict[int, float] | None])

• baseline_correlation (dict[int, float])

• cluster_label (dict[int, str | None])

• cluster_structure (dict[int, Literal['equicorrelated', 'toeplitz']])

• cluster_proxy_attenuation (dict[int, float])

• n_classes (int)

• class_names (list[str])

• samples_per_class (dict[int, int])

• class_sep (list[float])

• batch (BatchMeta | None)

• random_state (int | None)

• resolved_config (dict[str, object])

Return type:

None

Methods

__init__(feature_names, informative_idx, ...)
to_dict()	Convert to a plain dictionary (e.g., for JSON serialization).

Attributes

batch
batch_effects	Per-batch effect summary, or None if no batch effects were applied.
batch_labels	Per-sample batch assignments, or None if no batch effects were applied.
random_state
feature_names
informative_idx
noise_idx
corr_cluster_indices
anchor_idx
standalone_informative_groups
standalone_noise_range
mean_per_class_effect
covariance_per_class_correlation
baseline_correlation
cluster_label
cluster_structure
cluster_proxy_attenuation
n_classes
class_names
samples_per_class
class_sep
resolved_config

property batch_effects: ndarray | None

Per-batch effect summary, or None if no batch effects were applied.

Backward-compatibility accessor delegating to batch.

property batch_labels: ndarray | None

Per-sample batch assignments, or None if no batch effects were applied.

Backward-compatibility accessor delegating to batch.

to_dict()

Convert to a plain dictionary (e.g., for JSON serialization).

NumPy arrays are converted to plain lists so the result can be passed directly to json.dumps.

Return type:

dict[str, object]