https://github.com/douglasrizzo/pydunn

Dunn index in Python
https://github.com/douglasrizzo/pydunn

Last synced: 10 days ago
JSON representation

Dunn index in Python

• Host: GitHub
• URL: https://github.com/douglasrizzo/pydunn
• Owner: douglasrizzo
• License: mit
• Created: 2024-03-16T22:57:08.000Z (10 months ago)
• Default Branch: master
• Last Pushed: 2024-08-08T18:18:44.000Z (5 months ago)
• Last Synced: 2024-12-17T19:28:30.296Z (23 days ago)
• Language: Python
• Size: 33.2 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Dunn index in Python

An implementation of the Dunn index for internal cluster validity in Python. It sat for ages on a GitHub Gist but now it's been transferred to a proper repo.

## Usage

```py

import numpy as np

from sklearn.metrics.pairwise import euclidean_distances

from pydunn import dunn

# data points and labels

data = np.array([[0, 0], [0, 1], [1, 0], [1, 1], [10, 10], [10, 14], [14, 10], [14, 14]])

labels = [0, 0, 0, 0, 1, 1, 1, 1]

distances = euclidean_distances(data)

# compute the Dunn index

print("\n\n#### Dunn ####")

for diameter_method in DiameterMethod:

    for cdist_method in ClusterDistanceMethod:

        dunn_index = dunn(labels, distances, diameter_method, cdist_method)

        print(diameter_method, cdist_method, dunn_index)

```

As a bonus, you can also compute inter-cluster distances and cluster diameters separately.

```py

from pydunn import inter_cluster_distances, compute_cluster_diameters

# compute inter-cluster distances

print("#### Distances ####")

for cdist_method in ClusterDistanceMethod:

    print(cdist_method, "\n", inter_cluster_distances(labels, distances, cdist_method))

# compute cluster diameters

print("\n\n#### Diameters ####")

for diameter_method in DiameterMethod:

    print(diameter_method, compute_cluster_diameters(labels, distances, diameter_method))

```