https://github.com/yhenon/pyrcc

Python implementation of Robust Continuous Clustering
https://github.com/yhenon/pyrcc

Last synced: 3 months ago
JSON representation

Python implementation of Robust Continuous Clustering

• Host: GitHub
• URL: https://github.com/yhenon/pyrcc
• Owner: yhenon
• License: mit
• Created: 2017-09-13T21:24:07.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2019-07-08T16:05:29.000Z (about 6 years ago)
• Last Synced: 2024-11-04T17:47:16.570Z (8 months ago)
• Language: Python
• Size: 854 KB
• Stars: 101
• Watchers: 7
• Forks: 25
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# pyrcc

A python implementation of Robust Continuous Clustering.

The original matlab implementation can be found [here](https://bitbucket.org/sohilas/robust-continuous-clustering).

Sklearn style demonstration:

![rcc_clustering](imgs/rcc_image_comparison.png?raw=true)

RCC is a clustering method introduced here: http://www.pnas.org/content/early/2017/08/28/1700770114

This is a port of the matlab implementation provided by the authors.

The code is self-contained in rcc.py

The following parameters are used in RCC:

- `k`: (int)(deafult `10`) number of neighbors used in the mutual KNN graph

- `verbose`: (bool)(default `True`) verbosity 

- `preprocessing`: (string)(default "none") one of 'scale', 'minmax', 'normalization', 'none'. How to preprocess the features

- `measure`: (string)(default "euclidean") one of 'cosine' or 'euclidean'. Paper used 'cosine'. Metric to use in constructing the mutual KNN graph

- `clustering_threshold`: (float)(default 1.0) controls how agressively to assign points to clusters.

A demonstration of how to use this is shown in demo.py, measuring the AMI (adjusted mutual information) using the pendigits dataset.