Skip to content

Home

Jump to bottom
Thomas Bonald edited this page Apr 12, 2020 · 35 revisions

Scikit-Network is an open-source python package created at LINCS.

You are very welcome to participate and add your own graph algorithms.

We just ask you to follow some guidelines:

Allowed external dependencies

Coding style

  • Object oriented programming with Scikit-learn style: an algorithm is a class with a fit method.
  • PEP8 coding style (it is recommended to use Pycharm Community as an IDE)
  • Explicit inputs / outputs: adjacency, biadjacency, labels, scores, etc.

Documentation

labels: np.ndarray(dtype=int)
vector of labels

  • The description of each function or class should begin with a verb (imperative form) or a noun.

Cluster nodes of the graph.

  • The description of each function or class should include an example, in the following form:

imports
algo
data
output

  • Reference style: APA

Blondel, V. D., Guillaume, J. L., Lambiotte, R., & Lefebvre, E. (2008). Fast unfolding of communities in large networks. Journal of statistical mechanics: theory and experiment.

Tests

  • Each function or class must be tested.

Tutorials

  • The main modules should be illustrated by a Jupyter notebook.
Clone this wiki locally