Thorsten Wagner

@akbertram Do you have any idea how much an improved match() function would speed up the random forest predict function?

From my point it would be nice if it the improvements include float values as well :-) Just because all of my features in my project are float values :-)

Thank you for the improvement, this is great!! Here is a dump of my feature set: https://dl.dropboxusercontent.com/u/560426/training.csv.tar.gz

Thank you for your efforts :-)

I'm willing to help. I've done some research: This is an c implementation of sum_r8: https://searchcode.com/codesearch/view/63387344/ Here is the documentation of this function: https://gcc.gnu.org/onlinedocs/gfortran/SUM.html c implementation internal_pack: https://searchcode.com/codesearch/view/63387190/ c implementation...

Any progress here? :-)

Any progress here? At the moment this is really a bottle neck for me :-( Or are there any alternative to do it faster?

Thanks! This works perfectly!!!

Thanks for checking! Why do you want to apply the threshold on the difference? I would rather apply on the entries in the rotation matrix itself?

Another example:  Configuration: Sigma: 2.52 Lower Threshold: 2.21 Upper Threshold: 9.00