Potential bugs?: cna() with minus.log=TRUE
Issue #67
resolved
The input for the construction of clustered.cij is -log(abs(cij)). If the collaspe.method="max", this will take the weakest coupling between two communities as the coupling of coarse-grained network. Is it as expected?
Also, plot.cna() and view.cna() take edge weights as line width, which is opposite to the coupling strength. It may cause confusion that very strongly coupled two communities actually have unobservable edges.
Comments (3)
-
-
-
assigned issue to
-
assigned issue to
-
- changed status to resolved
We solved the bug related to -log(abs.cij) in cna and plot.cna function
- Log in to comment
Hmm, if we are using minus values here then the "max" option to collapse.matrix will be misleading as you describe.
We should fix this as ASAP by checking for -log(abs(cij)) input and then switching the option for collapse.matrix accordingly. Obviously, if we are using regular abs(cij) values then things are ok as is.
I will assign this to Guido but can also help if needed here.