random forest in LabVIEW

[EngrStudent]

A random forest is a universal function approximator (ufa), like a Neural Network.  UFA means that with enough training inputs it can approximate any reasonably behaved function to arbitrary precision.  Done right it is fast.  It is very fast.  It also does things like compute nonlinear variable importance without having to account directly for nonlinearity. 

MatLab has it as a specialized library.  R has it in a dozen flavors.  Python has it.  JMP has it.  It is of the same class of the winningest algorithms on kaggle: CART models.

Breiman also provided source code (Fortran)

https://www.stat.berkeley.edu/~breiman/RandomForests/cc_examples/prog.f

Here is something hacked using DLL's:

http://machine-vision-adhiguna.blogspot.com/2010/10/combining-intel-opencv-and-ni-labview.html

How to crash it using sklearn (sci-kit, R), LabPython

https://lavag.org/topic/20032-labpython-1050-error-shown-when-import-sklearn-library/

Content for homebrewing it in Python, or translating:

http://shop.oreilly.com/product/9780596529321.do

I really would like to get something clean/fast/prebuilt instead of translating Breiman's code to MathScript.  Does such a beastie exist? 

Is there a random forest for LabVIEW?

[Bob_Schor]

LabVIEW does have some neural network tools.  If you are interested in pursuing this Project, it makes a whole lot of sense (to me) to use the LabVIEW tools for neural networks instead of translating Fortran code for a neural network into MathScript code and then trying to run this model in LabVIEW.

Bob Schor

[EngrStudent]

@Bob - the Random Forest, while being a universal function approximator has some super-powers of its own against Neural Networks.  I can easily handle on a laptop the same tasks that if I did with a bleeding edge top-of-the-line workstation would take a compute-week.