README.md
In aydindemircioglu/yakmoR: A Simple Wrapper for the k-Means Library Yakmo

yakmoR

yakmoR is a simple wrapper for the K-Means C++ library (see www.tkl.iis.u-tokyo.ac.jp/~ynaga/yakmo/) developed by Naoki Yoshinaga.

yakmoR implements orthogonal K-Means. It can work in several rounds. In the first round, a normal K-Means is applied to the data. In each subsequent round, the next clustering is done on a subspace orthogonal to the centroids of the last clustering. This way one produces different views on the data. To speed up the whole procedure, Greg Hamerlys faster K-Means is utilized. Initilization can be done either classically (uniformly random) or by using the K-Means++ scheme.

Note

To circumvent platform-dependent floating point problems, there are some differences to the original yakmo package:

The different random number generators are replaced by the R::unif random number generator
The projection is rounded to 14 digits precision.

Therefore the results are not directly comparable to the original yakmo package. As the package is also enforced not to use specific C-compiler flags, compiling is NOT done with the -ffloat-store flag. This means now that different precision is used on different platforms (32bit vs 64bit, possibly also linux vs windows). So it is not really possible to build platform-independent packages using yakmoR. The workaround would be to enable the -ffloat-store flag. If you know a good way around, contact me.