DenMune#

class denmune_skl.DenMune(k_nearest=10, reduce_dims=True, target_dims=2, dim_reducer='tsne', metric='euclidean', metric_params=None, n_jobs=None, random_state=None)#

DenMune: A density-peak based clustering algorithm.

This implementation is a refactored and optimized version of the original algorithm published in Pattern Recognition (2021). It adheres to the scikit-learn API and offers significant performance improvements.

Parameters:

k_nearestint, default=10: The number of nearest neighbors to use for density estimation. This is the primary parameter of the algorithm.
reduce_dimsbool, default=True: If True, performs dimensionality reduction to target_dims before clustering. Recommended for high-dimensional data.
target_dimsint, default=2: The number of dimensions to reduce the data to if reduce_dims is True.
dim_reducerstr or estimator object, default=’tsne’: The dimensionality reduction method to use. Can be ‘tsne’, ‘pca’, or a pre-initialized scikit-learn compatible estimator object.
metricstr, default=’euclidean’: The distance metric to use for the k-nearest neighbor search. See sklearn.neighbors.NearestNeighbors for valid options.
metric_paramsdict, default=None: Additional keyword arguments for the metric function.
n_jobsint, default=None: The number of parallel jobs to run for neighbors search and dimensionality reduction. None means 1, -1 means using all processors.

Attributes:

labels_np.ndarray of shape (n_samples,): Cluster labels for each point in the dataset given to fit(). Noise points are labeled -1.
core_sample_indices_np.ndarray of shape (n_core_samples,): Indices of core samples. In the context of DenMune, these are the “Strong Points” that have an in-degree of at least k_nearest in the k-NN graph and are used to form the initial cluster skeletons.
n_clusters_int: The estimated number of clusters found by the algorithm. This does not include the noise cluster if one exists.
density_scores_np.ndarray of shape (n_samples,): The density score calculated for each point. In this implementation, this corresponds to the in-degree (|KNN_p<-|) from the k-NN graph, which is used for the “canonical ordering” step of the algorithm.
reducer_estimator object: The fitted dimensionality reduction estimator instance. This attribute is only set if the reduce_dims parameter is True and the number of input features is greater than target_dims. None if dimensionality reduction is not performed.
nn_NearestNeighbors: The fitted NearestNeighbors instance used to find the k-nearest neighbors of each point.
projected_X_np.ndarray of shape (n_samples, target_dims): The input data X after dimensionality reduction has been applied. If reduce_dims is False, or if dimensionality is not possible, this is a copy of the original X.
n_samples_int: The number of samples in the input data X.
n_features_in_int: The number of features seen during fit. Not available if metric == "precomputed".

References

[1] Abbas, M., El-Zoghabi, A., & Shoukry, A. (2021). DenMune: Density peak based: clustering using mutual nearest neighbors. Pattern Recognition, 109, 107589. https://doi.org/10.1016/j.patcog.2020.107589

Examples

>>> from sklearn.datasets import make_moons
>>> import numpy as np
>>> # Assuming DenMune is defined in the current scope or imported
>>> from denmune_skl import DenMune
>>>
>>> # Generate sample data
>>> X, y = make_moons(n_samples=200, noise=0.05, random_state=42)
>>>
>>> # Compute DenMune clustering
>>> model = DenMune(k_nearest=15, random_state=42)
>>> model.fit(X)
DenMune(k_nearest=15, random_state=42)
>>>
>>> # Access the results
>>> labels = model.labels_
>>> n_clusters = model.n_clusters_
>>> print(f"Estimated number of clusters: {n_clusters}")
Estimated number of clusters: 2

Methods

`fit`(X[, y])	Perform DenMune clustering.
`fit_predict`(X[, y])	Compute cluster centers and predict cluster index for each sample.
`get_metadata_routing`()	Get metadata routing of this object.
`get_params`([deep])	Get parameters for this estimator.
`set_params`(**params)	Set the parameters of this estimator.

fit(X, y=None)#: Perform DenMune clustering.

fit_predict(X, y=None)#

Compute cluster centers and predict cluster index for each sample.

Convenience method; equivalent to calling fit(X) followed by accessing the labels_ attribute.

Parameters:

Xarray-like of shape (n_samples, n_features): New data to transform.
yIgnored: Not used, present here for API consistency by convention.

Returns:

labelsndarray of shape (n_samples,): Index of the cluster each sample belongs to.

get_metadata_routing()#

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:

routingMetadataRequest: A MetadataRequest encapsulating routing information.

get_params(deep=True)#

Get parameters for this estimator.

Parameters:

deepbool, default=True: If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

paramsdict: Parameter names mapped to their values.

set_params(**params)#

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters: