help
help > RE: Seeking to understand “cost” and its effect on network properties
Jul 11, 2022 05:07 PM | Alfonso Nieto-Castanon - Boston University
RE: Seeking to understand “cost” and its effect on network properties
Hi John,
Yes, exactly, network-level cost is just the proportion of edges in a network (among all possible edges = # of pairs of nodes), so a value of 0.15 means that we are creating a graph by considering the network of all ROIs and keeping the 15% strongest connections among them. See https://web.conn-toolbox.org/fmri-methods/connectivity-measures/graphs-roi-level for a few additional details and references
Best
Alfonso
Originally posted by John Johnson:
Yes, exactly, network-level cost is just the proportion of edges in a network (among all possible edges = # of pairs of nodes), so a value of 0.15 means that we are creating a graph by considering the network of all ROIs and keeping the 15% strongest connections among them. See https://web.conn-toolbox.org/fmri-methods/connectivity-measures/graphs-roi-level for a few additional details and references
Best
Alfonso
Originally posted by John Johnson:
From the docs:
So I assume when Cost is set to 0.15, it thresholding the adjacency matrix so that the top 15% of the connections remain?
Adjacency matrix thresholding is typically
implemented using a fixed network cost level (e.g. keeping the
strongest 10% of connections) in order to allow sensitive
between-network comparisons of other graph measures of
interest
So I assume when Cost is set to 0.15, it thresholding the adjacency matrix so that the top 15% of the connections remain?
Threaded View
| Title | Author | Date |
|---|---|---|
| John Johnson | Jul 8, 2022 | |
| John Johnson | Jul 10, 2022 | |
| Alfonso Nieto-Castanon | Jul 11, 2022 | |
| John Johnson | Jul 11, 2022 | |