Analyze Network

Users may analyze preexisting networks that have been uploaded publicly by the UMCD and other users. The UMCD offers two analytical approaches:
Analyzing a network allows you to run a quick analysis on any publicly shared network. After making your selections in the fields below and clicking the Analyze button, the following information will be presented:

GeneralConnectivity
  • Basic Network Information
  • Global Network Metrics
    • Small World Attributes
    • Edge weight attributes
  • 3D Network
  • Connectivity Matrix
GraphsPlots
  • Node Degree Distribution
  • Node Degree Measures
  • Note Betweenness Centrality Bar Graph
  • Node Clustering Coefficient Bar Graph
  • Node Regional Efficiency Bar Graph
  • Node Participation Coefficient Bar Graph
  • Network Degree Plot
  • Network Betweenness Centrality Plot
  • Network Clustering Coefficient Plot
  • Network Regional Efficiency Plot
  • Network Participation Coefficient Plot
  • Modules Brain Plot
  • Modules Spring Plot

You can also view the regional report, download regional measures as a .txt file or download the entire report as a PDF.

The calculation of all graph theory metrics are based on their implementations in NetworkX. The following metrics are calculated:
  • Degree
  • Clustering Coefficient
  • Betweenness Centrality
  • Regional Efficiency
  • Participation Coefficient
  • Characteristic Path Length
  • Small Worldness*
  • Gamma**
  • Lambda***
  • Sigma (gamma/lambda)
  • Modularity (determined using the Louvain method with this program)
  • Spring layout

* Small Worldness attributes gamma (normalized clustering coefficient) and lambda (normalized path length) are calculated with respect to a population of random networks. Real networks are randomly rewired by 5000 double edge swaps . The random networks have the same degree distribution as the real networks and are connected. To minimize computation, only 10 random networks are constructed.
** Gamma is the ratio of the real network clustering coefficient to the mean of the random networks' clustering coefficients: cc_real / cc_random
*** Lambda is the ratio of the real network characteristic path length to the mean of the random networks' characteristic path lengths: cpl_real / cpl_random

Additional Analysis Details

  • For matrices uploaded with negative values, the weights are shifted such that the negative weights become the weakest positive weights.
  • For matrices uploaded with non-numeric values (inf or NaN), those values are set to 0.