pub fn make_sparse_classification(
config: &SparseClassConfig,
) -> SparseClassDatasetExpand description
Generate a high-dimensional sparse classification dataset.
Only n_informative out of n_features dimensions carry signal. Each
class gets a centroid drawn from N(0,1) projected only onto the informative
features. Samples are drawn by adding small Gaussian noise to their class
centroid; non-informative features remain exactly 0.
§Arguments
config- Generator configuration
§Returns
A SparseClassDataset where x has shape (n_samples × n_features).