[Eeglablist] STUDY cluster quality

Fran Copelli fcopelli at ryerson.ca
Tue Jan 12 06:38:00 PST 2021


Dear list,

1. My impression from the clustering function is that the *order* that the
clusters are formed is random. For example, if I recluster a STUDY, the
same clusters will appear again but in different order, which suggests to
me that they are listed in a random order every time. Can anyone confirm
this?

2. In one of my studies, my colleagues and I have noted how some of the
clusters have their ICs tightly-knit together (i.e., very close to the
center of the cluster), whereas others have ICs that appear to be very far
away from the center. Is there some way to evaluate the performance of a
cluster as a whole? Surely the more tightly-knit cluster is more likely to
be an actual source than the scattered cluster ...?

Thanks in advance!

Fran
Ryerson University, SMART Lab


More information about the eeglablist mailing list