Skip to content

zurk/lightcluster

 
 

Repository files navigation

lightcluster

Clustering lib (Python) by PreMoLab

  1. Install all necessary libraries:
  • 'igraph' (igraph.org) in 'lib' pip python_igraph-0.7.1.post6-cp27-none-win_amd64.whl install
  • 'networkx' (networkx.github.io)
  • 'scikit-learn' (scikit-learn.org)
  • 'agglomcluster' (pypi.python.org/pypi/AgglomCluster/1.0.2) in 'lib' python setup.py install
  • 'scan_by_enjoylife' will work automatically
  1. Use 'testing(multiple).py' or 'testing(single).py'. Follow the instructions given there.

  2. Available algorithms:

  • 'Spectral' from sklearn, described in 'articles\spectral_clustering'. Parameters: 'n_clusters'.
  • 'SCAN' by enjoylife, described in 'articles\scan'. Parameters: 'neighbours_threshold', 'similarity_threshold'.
  • 'GreedyNewman' from networkx, described in 'articles\newman_greedy'. Parameters: no.
  • 'Walktrap' from igraph, described in 'articles\walktrap'. Parameters: 'n_steps' --- length of random walks.
  • 'LPA' from igraph, described in 'articles\lpa'. Parameters: no.
  • 'CFinder' from stackoverflow.com/questions/20063927/overlapping- community-detection-with-igraph-or-other-libaries, described in 'articles\CFinder'. Parameters: 'clique_size' --- size of cliques
  • 'ClausetNewman' from igraph, described in 'articles\clauset-newman-moore' Parameters: no
  • 'Bigclam' from snap.stanford.edu, described in \articles\bigclam' Parameters: 'n_clusters'
  1. Available datasets: 'football.txt', 'polbooks.txt', 'protein_new.txt', 'scientists_new.txt', 'karate.txt', 'facebook.txt', 'cliques.txt', 'nested.txt', 'stars.txt', 'cycles.txt'.

  2. Available metrics: modularity, overlapping modularity, ratio cut, normalized cut, average F1-score, average recall, average precision, normalized mutual information (NMI), adjusted_rand_score (ARS)

About

Clustering lib by PreMoLab

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%