clustlib 0.1.0

Constrained clustering library for Python

Clustlib: Constrained Clustering Library

[!NOTE] 🚧 Currently the project is a Work in progress, but you can help us to fix it.

Please refer to our contributing documenation to know how can you start helping us

[!WARNING] This Library is a Work in progress at the moment, most use cases may not be contempleted or supported yet. If you find something feel free to create a bug report.

In the same way if you think we are missing something important create a feature request

What it is?

Clustlib is a Python library that provides a unified framework for constrained clustering algorithms.

Unlike traditional clustering, where groups are formed purely from data similarity, constrained clustering integrates prior knowledge in the form of must-link and cannot-link constraints.

This library includes implementations of both classic algorithms and state-of-the-art metaheuristics, enabling researchers and practitioners to experiment, compare, and extend methods in a common environment.

If you want to know more deep in our documentation

✨ Key Features

• Unified Python API across all algorithms
• Support for must-link and cannot-link constraints
• Scikit-learn inspired design for easy integration
• Implementations from different families:
  • K-Means based methods
  • Bayesian models
  • Evolutionary and metaheuristic approaches
• Modular and extensible codebase

🚀 Installation

pip install clustlib

Basic Usage

from clustlib.kmeans import COPKMeans

constraints = np.zeros((X.shape[0], X.shape[0]))

# Example: COP-KMeans
model = COPKMeans(n_clusters=3, constraints=constraints)
model.fit(X)
labels = model.labels_