Module datasets 

Dataset generators and loaders (Pillar 1: beat scikit-learn).

Mirrors sklearn.datasets. Provides synthetic generators (make_blobs, make_regression) and embedded real datasets (load_iris). The embedded data is sourced once from scikit-learn and committed, so loading has no runtime Python/network dependency. Larger embedded sets (load_digits, load_california_housing) are a follow-up (PMAT-720 continuation).

All generators are deterministic given a seed — the same seed yields byte-identical output, so benchmarks and falsifiers are reproducible.

Functions

load_iris

Load the classic Iris dataset: 150 samples, 4 features (sepal length/width, petal length/width), 3 balanced classes (50 each).

make_blobs

Generate isotropic Gaussian blobs for clustering/classification.

make_classification

Generate a random n_classes-class classification problem.

make_regression

Generate a random linear regression problem with Gaussian noise.