Reputation: 131
There probably is name for the process or method I am looking for, but have not been able to find it. I have described it as best I could below.
I am looking for a process that reorders a symmetric matrix in terms of clusters, for example:
In order words it tries to group the indices by some measure. So if you were to split up the resulting matrix in sub-matrices, in the example above A and D are as 'positive' as possible while B and C are as 'negative' as possible.
Is there a name for this process? What would a computationally efficient approach be to implement this? (I plan to use it for quite large structures in Python). Any pointers very welcome.
Upvotes: 3
Views: 411
Reputation: 77505
Use e.g. single-linkage clustering.
This is commonly used to reorder matrixes.
In biclustering, this is often applied to both rows and columns; here you would only compute it once and apply it to both.
Upvotes: 3