Hello! I find your work interesting, and would like to apply for it. Here is my proposal. First, I've worked with CART family of algos as well as Naive Bayes ones using R, actually. If you know, R is a very powerful set of tools for machine learning using third party packages all available as OpenSOurce, you know right, no one alone can make the work great, so they open source it.
Anyways, I can get you job done in R. Also, I've worked with Weka in Matlab, alone. But, if the platform were not a mandatory I'd prefer R. If you like to be in Weka, we could do that too. What about stuff like Python? So, I've worked with all the 3 you've mentioned, especially K-means that I've been working a lot, and coded from scratch for assignments. I'm interested to work on the data mining on the RAW data. Do you use any outlier detection or just use it? Also,since it is supervised, I'll be expecting the data to be labelled. I do not have access to mTurk for batching people to get the job done.
Let me know when you're available for discussion. Else, we could just begin with the 3 ML routines. By the way, what changes you want to do, let me know that too. We could create a lib just for that, for future requirement.
Thanks!! Have a great day!