Skip to content

Networkx based Package for "pruning" weighted complex networks based on the Marginal Likelihood Filter.

License

Notifications You must be signed in to change notification settings

nikhilkanta/GraphPruning2

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GraphPruning2



Networkx based Package for "pruning" weighted complex networks based on the Marginal Likelihood Filter.

Overview

When dealing with weighted complex networks of dyadic relationships between nodes, we frequently encounter overly-dense "hairball" networks where the large number of edges may obfuscate the most important structures within the graph.

People often "prune" the graph by simply setting a threshold on the edge weights and removing all edges below the threshold. However, this simplistic approach systematically disfavours low-degree nodes and the structures they represent.Therefore, we need to measure the significance of an edge, with respect to the degrees of its incident vertices.

Here, we do this using a version of the configuration model as our null model that defines what we expect an edge’s weight to be given its end-nodes’ degrees. The null model predicts a probability distribution for the value of the edge weight. By comparing the observed weight of an edge with this predicted distribution, we can then compute a p-value which tells us how surprising the observed value is given the null model. This p-value gives us a measure of the statistical significance of the edge, and we can filter the edges according to this significance rather than the raw weight itself.

About

Networkx based Package for "pruning" weighted complex networks based on the Marginal Likelihood Filter.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Languages