I agree, it's not perfect - just a useful abstraction. Just the same as arbitrary thresholds for correlation or a p<0.05 significance level - often you lose information but gain insight. From personal experience I've…
Nice article. You could consider maximum spanning trees as a way to prune your correlation graph; they are very effective at suggesting underlying structure or kinetics of a system. Just use the minimum spanning tree…
I agree, it's not perfect - just a useful abstraction. Just the same as arbitrary thresholds for correlation or a p<0.05 significance level - often you lose information but gain insight. From personal experience I've…
Nice article. You could consider maximum spanning trees as a way to prune your correlation graph; they are very effective at suggesting underlying structure or kinetics of a system. Just use the minimum spanning tree…