Mapper on BallMapper

class pyballmapper.mobm.MapperonBallMapper(cover_BM: BallMapper, target_space: ndarray[tuple[Any, ...], dtype[_ScalarT]], eps: float, min_samples: int = 1, dbscan_metric: str = 'euclidean', sparse: bool = False, dbg: bool = False)

Bases: BallMapper

Mapper on BallMapper using DBSCAN as clustering algorithm.

It uses scipy csr sparse matrix to speed up computations.

Parameters:

  • cover_B (BallMapper) – Ball Mapper graph covering the image space.

  • target_space (array-like of shape (n_samples, n_features)) – The high dimensional pointcloud where to pull back from nodes of the BallMapper.

  • eps (float) – eps parameter for DBSCAN

  • min_samples (int) – min number of elements in a cluster that make it a cluster and not noise. Another DBSCAN parameter.

Graph

The resulting Mapper on BallMapper graph. With node attributes size and points covered.

Type:

NetworkX graph

Notes

https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html