I spent this week working to implement our algorithm in Weka. This is very difficult, since the Weka code is structured in a way that is not intuitive to our algorithm. Much of the information that the algorithm needs is never stored.
I spent a lot of time futilely tracking through the code in search of the the information I need. Once I realized it was not there and I would have to make many changes to the existing code, I was able to move forward very quickly. I now have working implementations of the kernel density part of the algorithm.