Hi,

I had a real world gene network (with scale free distribution) that I clustered into eight parts. I further checked the degree distribution of these eight clusters to check if it conformed to a power law distribution. I have an advisory committee and one of my committee members ended up asking me if that was necessary and that clustering algorithms do you take away the real world properties of a network. I couldn't answer him with much confidence and I have been stumped since then.

I checked for papers where clustering has been done but I could not see any that check for real world properties of the clusters after clustering. I am thinking that I will change my reasoning to also say that I wanted to study the rate of decay between the slopes of each cluster to form a comparison but I cannot find a study where decay constant between clusters have been compared as well. Kindly offer some guidance.