gcomvkm 0.1.0

Globally Collaborative Multi-View k-Means Clustering Algorithm

G-CoMVKM

Globally Collaborative Multi-View k-Means Clustering

Overview

G-CoMVKM is a Python implementation of the Globally Collaborative Multi-View k-Means clustering algorithm, originally developed by Kristina P. Sinaga. The algorithm integrates a collaborative transfer learning framework with entropy-regularized feature-view reduction, enabling dynamic elimination of uninformative components. This method achieves clustering by balancing local view importance and global consensus.

Installation

You can install G-CoMVKM directly from PyPI:

pip install gcomvkm

Requirements

• Python 3.7+
• NumPy
• SciPy
• Matplotlib
• scikit-learn
• seaborn

Usage

from gcomvkm import GCoMVKM
from gcomvkm.utils import load_synthetic_data
from gcomvkm.evaluation import nmi, rand_index, adjusted_rand_index

# Load the dataset
X, label = load_synthetic_data()

# Create and fit the model
model = GCoMVKM(
    n_clusters=2,
    gamma=5.0,
    theta=4.0,
    max_iter=100,
    tol=1e-4,
    verbose=True,
    random_state=42
)

model.fit(X)

# Get the cluster assignments
predicted_labels = model.labels_

# Evaluate the clustering performance
nmi_score = nmi(label, predicted_labels)
ri_score = rand_index(label, predicted_labels)
ari_score = adjusted_rand_index(label, predicted_labels)

print(f"NMI: {nmi_score:.4f}")
print(f"RI: {ri_score:.4f}")
print(f"ARI: {ari_score:.4f}")

Algorithm Parameters

• n_clusters: Number of clusters to form
• gamma: Exponent parameter to control the weights of V (typically in range [0,1])
• theta: Coefficient parameter to control the weights of W (typically > 0)
• max_iter: Maximum number of iterations
• tol: Convergence tolerance
• verbose: Whether to print progress information
• random_state: Random seed for reproducibility

Features

• Multi-view clustering with feature and view weighting
• Automatic dimensionality reduction through entropy regularization
• Balanced local and global consensus clustering
• Comprehensive evaluation metrics

Citation

If you use this package in your research, please cite the original paper:

@article{sinaga2025gcomvkm,
  title={A Globally Collaborative Multi-View k-Means Clustering},
  author={Sinaga, Kristina P.},
  journal={MDPI Electronics},
  year={2025}
}

License

This project is licensed under the MIT License - see the LICENSE file for details. 6. Performance metrics across different initializations 7. Convergence plot

References

This implementation is based on the MATLAB code by Kristina P. Sinaga. For more details about the algorithm, please refer to the original paper.