qbiocode.data_generation.make_blobs module#

Generate synthetic blob (Gaussian cluster) datasets.

This module creates multiple configurations of blob datasets with varying numbers of samples, features, centers, and cluster standard deviations, useful for testing clustering and classification algorithms.

Summary#

Functions:

`generate_blobs_datasets`	Generate multiple blob (Gaussian cluster) datasets with varying parameters.
`generate_default_blobs_datasets`	Generate blob datasets with default parameter configurations.

Reference#

generate_blobs_datasets(n_samples, n_features, centers, cluster_std, save_path=None, random_state=42)[source]#

Generate multiple blob (Gaussian cluster) datasets with varying parameters.

Creates a series of synthetic datasets consisting of isotropic Gaussian blobs for clustering and classification tasks. Each configuration varies the number of samples, features, cluster centers, and cluster spread.

Parameters:

n_samples (list of int) – List of sample sizes to generate for each configuration. Example: [100, 200, 300]
n_features (list of int) – List of feature dimensions to generate. Example: [2, 4, 8]
centers (list of int) – List of numbers of cluster centers (classes). Example: [2, 3, 4]
cluster_std (list of float) – List of standard deviations of the clusters. Example: [0.5, 1.0, 1.5, 2.0]
save_path (str, optional) – Directory path to save generated datasets. If None, datasets are not saved. Default: None
random_state (int, optional) – Random seed for reproducibility. Default: 42

Returns:

Dictionary containing generated datasets with keys as configuration strings and values as (X, y) tuples, where

X is the feature matrix, a pd.DataFrame of shape (n_samples, n_features)
y holds the target labels, a pd.Series of shape (n_samples,)

Return type:

dict

Notes

Generates all combinations of input parameters
Each blob is an isotropic Gaussian distribution
Useful for testing classification and clustering algorithms
Blobs are well-separated when cluster_std is small relative to center distances

Examples

>>> from qbiocode.data_generation import generate_blobs_datasets
>>>
>>> # Generate simple blob datasets
>>> datasets = generate_blobs_datasets(
...     n_samples=[100, 200],
...     n_features=[2, 4],
...     centers=[2, 3],
...     cluster_std=[1.0, 1.5]
... )
>>>
>>> # Access a specific configuration
>>> X, y = datasets['n_samples_100_n_features_2_centers_2_cluster_std_1.0']
>>> print(f"Shape: {X.shape}, Classes: {y.nunique()}")
Shape: (100, 2), Classes: 2

>>> # Save datasets to disk
>>> datasets = generate_blobs_datasets(
...     n_samples=[100],
...     n_features=[2],
...     centers=[3],
...     cluster_std=[1.0],
...     save_path='./data/blobs'
... )

See also

generate_circles_datasets: Generate concentric circles
generate_moons_datasets: Generate interleaving half-circles
generate_classification_datasets: Generate high-dimensional classification data

References

Pedregosa et al., “Scikit-learn: Machine Learning in Python”, JMLR 12, pp. 2825-2830, 2011.

generate_default_blobs_datasets(save_path=None, random_state=42)[source]#

Generate blob datasets with default parameter configurations.

Convenience function that generates a standard set of blob datasets using predefined parameter ranges suitable for most testing scenarios.