cknn package

Submodules

cknn.cknn module

class cknn.cknn.CkNearestNeighbors(n_neighbors=5, delta=1.0, metric='euclidean', t='inf', include_self=False, is_sparse=True)

Bases: object

This object provides the all logic of CkNN.

Parameters

• n_neighbors – int, optional, default=5 Number of neighbors to estimate the density around the point. It appeared as a parameter k in the paper.

• delta – float, optional, default=1.0 A parameter to decide the radius for each points. The combination radius increases in proportion to this parameter.

• metric – str, optional, default=’euclidean’ The metric of each points. This parameter depends on the parameter metric of scipy.spatial.distance.pdist.

• t –

  ‘inf’ or float or int, optional, default=’inf’ The decay parameter of heat kernel. The weights are calculated as follow:

  W_{ij} = exp(-(||x_{i}-x_{j}||^2)/t)

  For more infomation, read the paper ‘Laplacian Eigenmaps for Dimensionality Reduction and Data Representation’, Belkin, et. al.

• include_self – bool, optional, default=True All diagonal elements are 1.0 if this parameter is True.

• is_sparse – bool, optional, default=True The method cknneighbors_graph returns csr_matrix object if this parameter is True else returns ndarray object.

cknneighbors_graph(X)

A method to calculate the CkNN graph

Parameters

X – ndarray The data matrix.

return: csr_matrix (if self.is_sparse is True)

or ndarray(if self.is_sparse is False)

cknn.cknn.cknneighbors_graph(X, n_neighbors, delta=1.0, metric='euclidean', t='inf', include_self=False, is_sparse=True, return_instance=False)

Module contents

The cknn module implements the Continuous k-Nearest Neighbors[1].

Reference

1

T. Berry and T. Sauer, “Consistent man-ifold representation for topological dataanalysis,” 2016.

cknn.cknneighbors_graph(X, n_neighbors, delta=1.0, metric='euclidean', t='inf', include_self=False, is_sparse=True, return_instance=False)

class cknn.CkNearestNeighbors(n_neighbors=5, delta=1.0, metric='euclidean', t='inf', include_self=False, is_sparse=True)

Bases: object

This object provides the all logic of CkNN.

Parameters

• n_neighbors – int, optional, default=5 Number of neighbors to estimate the density around the point. It appeared as a parameter k in the paper.

• delta – float, optional, default=1.0 A parameter to decide the radius for each points. The combination radius increases in proportion to this parameter.

• metric – str, optional, default=’euclidean’ The metric of each points. This parameter depends on the parameter metric of scipy.spatial.distance.pdist.

• t –

  ‘inf’ or float or int, optional, default=’inf’ The decay parameter of heat kernel. The weights are calculated as follow:

  W_{ij} = exp(-(||x_{i}-x_{j}||^2)/t)

  For more infomation, read the paper ‘Laplacian Eigenmaps for Dimensionality Reduction and Data Representation’, Belkin, et. al.

• include_self – bool, optional, default=True All diagonal elements are 1.0 if this parameter is True.

• is_sparse – bool, optional, default=True The method cknneighbors_graph returns csr_matrix object if this parameter is True else returns ndarray object.

cknneighbors_graph(X)

A method to calculate the CkNN graph

Parameters

X – ndarray The data matrix.

return: csr_matrix (if self.is_sparse is True)

or ndarray(if self.is_sparse is False)