clustering

SpectralClustering

Bases: Clustering

assign_labels = 'discretize' instance-attribute

n_init = 100 instance-attribute

number_of_clusters = None instance-attribute

__init__(cycles)

cluster()

Cluster cycles using spectral clustering.

Returns:

Name	Type	Description
dict	ndarray	dictionary with cluster labels as keys, and all cycles with said label as values.

get_adjacency_matrix()

Build an adjacency matrix from a weighted graph created by the set of cycles that is being clustered. This graph contains all unique nodes and edges from all cycles combined. The weights of the edges are the number of times that each edge occurs in the set of cycles.

Returns:

Type	Description
ndarray	np.ndarray: adjacency matrix

get_eigengap(laplacian_matrix)

Calcualte the eigengap of the clusters. The eigengap is defined as the place where the gap between eigenvalues of the laplacian matrix is largest. The number of eigenvalues beneeth this gap define the number of clusters used for spectral clustering.

get_norm_laplacian(adjacency_matrix)

Calculates the normalized laplacian of the adjacency matrix.

Parameters:

Name	Type	Description	Default
adjacency_matrix	ndarray	adjacency matrix of the graph	required

set_assign_labels(assign_labels)

Default = 'discretize'

set_n_init(n_init)

Number of times the Number of time the k-means algorithm will be run with different centroid seeds. The final results will be the best output of n_init consecutive runs in terms of inertia.

Default = 100

set_number_of_clusters(number_of_clusters)

By default, the number of clusters is automatically calculated by analysing where the gap between eigenvalues lies.