API Reference / datasets / generators / clustering /

swiss_roll.py

Swiss roll generator: a 2-D sheet rolled through 3-D space.

Samples points from a plane that has been curled into a spiral, optionally with a hole punched through it. Points that are close in 3-D can be far apart along the sheet, which is what makes it the standard test for manifold learning and dimensionality reduction.

Classes

SwissRoll

class datasets.generators.clustering.swiss_roll.SwissRoll(ClusteringGenerator)

Swiss Roll data generator.

Generates the classic Swiss Roll 3D manifold dataset. This is commonly used to test dimensionality reduction algorithms.
Constructor 
__init__(
    self,
    n_samples: int = 100,
    noise: float = 0.0,
    hole: bool = False,
    random_state: Optional[int] = None,
)

Parameters

n_samples
int = 100
Number of samples to generate.
noise
float = 0.0
Standard deviation of Gaussian noise.
hole
bool = False
Whether to include a hole in the middle.
random_state
int or None = None
Random seed for reproducibility.
python 
>>> from tuiml.datasets.generators.clustering import SwissRoll
>>> gen = SwissRoll(n_samples=1000, noise=0.1, random_state=0)
>>> data = gen.generate()
>>> data.X.shape                 # always 3-D
(1000, 3)

Methods

generate (self) -> GeneratedData

Generate Swiss Roll data.

Returns
GeneratedData
Generated dataset with feature array X of shape (n_samples, 3) and continuous position values y.
get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.