Apply clustering to a projection of the normalized Laplacian.
In practice Spectral Clustering is very useful when the structure of the individual clusters is highly non-convex, or more generally when a measure of the center and spread of the cluster is not a suitable description of the complete cluster, such as when clusters are nested circles on the 2D plane.
If the affinity matrix is the adjacency matrix of a graph, this method can be used to find normalized graph cuts [1], [2].
When calling fit, an affinity matrix is constructed using either a kernel function such the Gaussian (aka RBF) kernel with Euclidean distance d(X, X):
new SpectralClustering(opts?: object): SpectralClustering;| Name | Type | Description |
|---|---|---|
opts? |
object |
- |
opts.affinity? |
string |
‘nearest_neighbors’: construct the affinity matrix by computing a graph of nearest neighbors. Default Value 'rbf' |
opts.assign_labels? |
"kmeans" | "discretize" | "cluster_qr" |
The strategy for assigning labels in the embedding space. There are two ways to assign labels after the Laplacian embedding. k-means is a popular choice, but it can be sensitive to initialization. Discretization is another approach which is less sensitive to random initialization [3]. The cluster_qr method [5] directly extract clusters from eigenvectors in spectral clustering. In contrast to k-means and discretization, cluster_qr has no tuning parameters and runs no iterations, yet may outperform k-means and discretization in terms of both quality and speed. Default Value 'kmeans' |
opts.coef0? |
number |
Zero coefficient for polynomial and sigmoid kernels. Ignored by other kernels. Default Value 1 |
opts.degree? |
number |
Degree of the polynomial kernel. Ignored by other kernels. Default Value 3 |
opts.eigen_solver? |
"arpack" | "lobpcg" | "amg" |
The eigenvalue decomposition strategy to use. AMG requires pyamg to be installed. It can be faster on very large, sparse problems, but may also lead to instabilities. If undefined, then 'arpack' is used. See [4] for more details regarding 'lobpcg'. |
opts.eigen_tol? |
number |
Stopping criterion for eigen decomposition of the Laplacian matrix. If eigen\_tol="auto" then the passed tolerance will depend on the eigen\_solver: Default Value 'auto' |
opts.gamma? |
number |
Kernel coefficient for rbf, poly, sigmoid, laplacian and chi2 kernels. Ignored for affinity='nearest\_neighbors'. Default Value 1 |
opts.kernel_params? |
any |
Parameters (keyword arguments) and values for kernel passed as callable object. Ignored by other kernels. |
opts.n_clusters? |
number |
The dimension of the projection subspace. Default Value 8 |
opts.n_components? |
number |
Number of eigenvectors to use for the spectral embedding. If undefined, defaults to n\_clusters. |
opts.n_init? |
number |
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. Only used if assign\_labels='kmeans'. Default Value 10 |
opts.n_jobs? |
number |
The number of parallel jobs to run when affinity='nearest\_neighbors' or affinity='precomputed\_nearest\_neighbors'. The neighbors search will be done in parallel. undefined means 1 unless in a joblib.parallel\_backend context. \-1 means using all processors. See Glossary for more details. |
opts.n_neighbors? |
number |
Number of neighbors to use when constructing the affinity matrix using the nearest neighbors method. Ignored for affinity='rbf'. Default Value 10 |
opts.random_state? |
number |
A pseudo random number generator used for the initialization of the lobpcg eigenvectors decomposition when eigen\_solver \== 'amg', and for the K-Means initialization. Use an int to make the results deterministic across calls (See Glossary). |
opts.verbose? |
boolean |
Verbosity mode. Default Value false |
Defined in: generated/cluster/SpectralClustering.ts:27
Disposes of the underlying Python resources.
Once dispose() is called, the instance is no longer usable.
dispose(): Promise<void>;Promise<void>
Defined in: generated/cluster/SpectralClustering.ts:200
Perform spectral clustering from features, or affinity matrix.
fit(opts: object): Promise<any>;| Name | Type | Description |
|---|---|---|
opts |
object |
- |
opts.X? |
ArrayLike |
Training instances to cluster, similarities / affinities between instances if affinity='precomputed', or distances between instances if affinity='precomputed\_nearest\_neighbors. If a sparse matrix is provided in a format other than csr\_matrix, csc\_matrix, or coo\_matrix, it will be converted into a sparse csr\_matrix. |
opts.y? |
any |
Not used, present here for API consistency by convention. |
Promise<any>
Defined in: generated/cluster/SpectralClustering.ts:217
Perform spectral clustering on X and return cluster labels.
fit_predict(opts: object): Promise<ArrayLike>;| Name | Type | Description |
|---|---|---|
opts |
object |
- |
opts.X? |
ArrayLike |
Training instances to cluster, similarities / affinities between instances if affinity='precomputed', or distances between instances if affinity='precomputed\_nearest\_neighbors. If a sparse matrix is provided in a format other than csr\_matrix, csc\_matrix, or coo\_matrix, it will be converted into a sparse csr\_matrix. |
opts.y? |
any |
Not used, present here for API consistency by convention. |
Promise<ArrayLike>
Defined in: generated/cluster/SpectralClustering.ts:257
Get metadata routing of this object.
Please check User Guide on how the routing mechanism works.
get_metadata_routing(opts: object): Promise<any>;| Name | Type | Description |
|---|---|---|
opts |
object |
- |
opts.routing? |
any |
A MetadataRequest encapsulating routing information. |
Promise<any>
Defined in: generated/cluster/SpectralClustering.ts:301
Initializes the underlying Python resources.
This instance is not usable until the Promise returned by init() resolves.
init(py: PythonBridge): Promise<void>;| Name | Type |
|---|---|
py |
PythonBridge |
Promise<void>
Defined in: generated/cluster/SpectralClustering.ts:140
boolean=false
Defined in: generated/cluster/SpectralClustering.ts:25
boolean=false
Defined in: generated/cluster/SpectralClustering.ts:24
PythonBridge
Defined in: generated/cluster/SpectralClustering.ts:23
string
Defined in: generated/cluster/SpectralClustering.ts:20
any
Defined in: generated/cluster/SpectralClustering.ts:21
Affinity matrix used for clustering. Available only after calling fit.
affinity_matrix_(): Promise<ArrayLike[]>;Promise<ArrayLike[]>
Defined in: generated/cluster/SpectralClustering.ts:339
Names of features seen during fit. Defined only when X has feature names that are all strings.
feature_names_in_(): Promise<ArrayLike>;Promise<ArrayLike>
Defined in: generated/cluster/SpectralClustering.ts:420
Labels of each point
labels_(): Promise<ArrayLike>;Promise<ArrayLike>
Defined in: generated/cluster/SpectralClustering.ts:366
Number of features seen during fit.
n_features_in_(): Promise<number>;Promise<number>
Defined in: generated/cluster/SpectralClustering.ts:393
py(): PythonBridge;PythonBridge
Defined in: generated/cluster/SpectralClustering.ts:127
py(pythonBridge: PythonBridge): void;| Name | Type |
|---|---|
pythonBridge |
PythonBridge |
void
Defined in: generated/cluster/SpectralClustering.ts:131