Fuzzy c-Means for Rapid Miner
I noticed that Rapid miner lacks of some clustering algorithms. Especially Fuzzy c-Means and its derivatives. Also I was experimenting with the other clustering algorithm models available and it seems there are not many of them. I was expecting many more. I don't know if there are plugins or extensions for clustering, I did not find them. In the Weka extension, there are also not so many clustering algorithms. If any one can point me to more of them, I would be very glad. I googled for it, but unfortunately, I did not find any.
I am a researcher in clustering and I have quite some algorithms developed. I want to use RM to publish my algorithms not only as a paper, but also its implementation. I already ordered the documentation but my company is rather slow for such things. Anyway, I noticed that the DBScan implementation is very weak. First, its very slow and Second, the result is wrong. I already filed a bug-report for that, but I have an implementation that is very fast. It has a execution complexity of n*log(n) for building a tree data structure based on the specified distance function and k*n*log(n) for executing DBScan it self. I don't know how fast the implementation is, but it needs for 35000 data objects of each 23 real values more than 2 hours. I will export the data set before clustering and apply my algorithm on it to see how fast it is. But from my experience with other (much larger) data sets, it should be done within a few seconds, maybe minutes. Is there any way to contribute to rapid miner and improve existing algorithms?