search

micans / mcl

MCL, the Markov Cluster algorithm, also known as Markov Clustering, is a method and program for clustering weighted or simple networks, a.k.a. graphs.
https://micans.org/mcl
Other
86 stars 12 forks source link

clm info output question #20

Closed jmayoral1 closed 8 months ago

jmayoral1 commented 8 months ago

Greetings,

Not an issue, but a question about the output from clm info: I understand that 'eff' is efficiency, 'mf' is mass fraction, and 'af' is area fraction, but I'm not sure what 'mod' is short for. Based on the values, I infer that this might be 'modularity', but is this correct? I couldn't find a description of 'mod' in the clm info documentation online. Apologies if I missed it somewhere. Below is an example of the output:

eff=0.69641 mod=0.99225 mf=0.99317 af=0.00044 src=out.seq.mci.I12 ncl=33952 max=2016 ctr=162.3 avg=10.9 min=1 DGI=2016 TWI=1067 TWL=99 sgl=19635 qrt=26608

I'm using mcl version 22-282.

Regards, Josh

micans commented 8 months ago

You are correct on both counts, it is missing in the documentation - I'll fix that, and it is modularity being measured.

Efficiency favours granular/fine-grained clusterings, modularity favours coarser clusterings. At some point I looked at e.g. adding the two up to see if that would give a nice parabolic maximum, but to no avail. Any single criterion (e.g. efficiency/modularity/conductance) will favour a certain granularity range. For more complex graphs that admit cluster structure at different levels I don't think any of them can be favoured, unless of course they happen to pick the granularity range that best suits your application.

jmayoral1 commented 8 months ago

Got it, thank you for the quick reply and explanation! Closing this "issue".