clusteval

clusteval is a python package that is developed to evaluate detected clusters and return the cluster labels that have most optimal clustering tendency, Number of clusters and clustering quality. Multiple evaluation strategies are implemented for the evaluation; silhouette, dbindex, and derivative, and four clustering methods can be used: agglomerative, kmeans, dbscan and hdbscan.

Blogs

1. A step-by-step guide for clustering images
2. Detection of Duplicate Images Using Image Hash Functions
3. From Data to Clusters: When is Your Clustering Good Enough?
4. From Clusters To Insights; The Next Step

Documentation

Full documentation is available at erdogant.github.io/clusteval, including examples and API references.

Installation

It is advisable to use a virtual environment:

conda create -n env_clusteval python=3.12
conda activate env_clusteval

Install via PyPI:

pip install clusteval

To upgrade to the latest version:

pip install --upgrade clusteval

Import the library:

from clusteval import clusteval

Examples

A structured overview is available in the documentation.

Silhouette Score	Optimal Clusters
Dendrogram	Davies-Bouldin Index
Derivative Method	DBSCAN
HDBSCAN A	HDBSCAN B

Citation

Please cite clusteval in your publications if it has been helpful in your research. Citation information is available at the top right of the GitHub page.

Related Tools & Blogs

• Use ARI when clustering contains large equal-sized clusters
• Use AMI for unbalanced clusters with small components
• Adjusted Rand Score — scikit-learn
• Adjusted for Chance Measures — scikit-learn
• imagededup GitHub repo
• Clustering images by visual similarity
• Facebook DeepCluster
• PCA on Hyperspectral Data
• Face Recognition with PCA

Star history

Contributors

Thank the contributors!

Maintainer

• Erdogan Taskesen, github: erdogant
• Contributions are welcome.
• Yes! This library is entirely free but it runs on coffee! :) Feel free to support with a Coffee.