src.model package

Module contents

class src.model.SpectralEmbedding(radius, n_components=2, eps=None, eigen_solver='amg', drop_first=True, lap_type='geometric', diffusion_time=0.0, diffusion_maps=False, **solver_kwds)

Bases: Embedding

Parameters:

  • radius (float)

  • n_components (int)

  • eps (float | None)

  • eigen_solver (Literal['dense', 'arpack', 'lobpcg', 'amg'])

  • drop_first (bool)

  • lap_type (Literal['geometric', 'random_walk', 'symmetric'])

  • diffusion_time (float)

  • diffusion_maps (bool)

  • solver_kwds (Any)

aff_mat: AffinityMatrix | None = None
dist_mat: DistanceMatrix | None = None
fit(data, save=False)

Fit the model from geometry matrix data.

Parameters:

  • data (Points | AffinityMatrix | DistanceMatrix | LaplacianMatrix) – Any GeometryType geometry matrix. If a LaplacianMatrix, overrides the lap_type specified in the constructor.

  • save (bool) – Whether to save the DistanceMatrix, AffinityMatrix and LaplacianMatrix. IMPORTANT: If True, will take a significant additional memory. If your dataset is a significant portion of your available memory, it is reccomended to keep save False.

Returns:

The EigenSystem corresponding to the spectral embedding.

Return type:

EmbeddingSystem

lap_mat: LaplacianMatrix | None = None