Multiclass RandomForest

[satyarth934]

Implementation of Multiclass RandomForest.

[phsft-bot]

Can one of the admins verify this patch?

[satyarth934]

May I please know as to what kinds of tests are these? It works fine on my machine.

[vgvassilev]

If you click on "Details" you will see. You should follow ROOT coding convention.

[satyarth934]

Hi @vgvassilev ,
Following are the kinds of additional lines that I see added to the original code when I clicked Details.

diff --git a/tmva/pymva/src/MethodPyRFOneVsRest.cxx b/tmva/pymva/src/MethodPyRFOneVsRest.cxx
index b1113d1..9b41b39 100644
--- a/tmva/pymva/src/MethodPyRFOneVsRest.cxx
+++ b/tmva/pymva/src/MethodPyRFOneVsRest.cxx
@@ -16,7 +16,7 @@

Similar lines are added all over.
Rest of the code looks more like an output of git diff.

I am unable to understand, how to interpret these additional added lines and how to understand my mistakes.

[vgvassilev]

@phsft-bot build!

@satyarth934, it seems you found your way around this. The idea of the diff is that you can apply it without having clang format. You just open the log file in plain text copy the diff, store it on your machine and apply it with git apply.

[satyarth934]

@vgvassilev , Thanks.
Looks fine now.

[omazapa]

Dear @satyarth934
This implementation is not under the architecture and design of TMVA, you dont need to create a new class to add support for multiclass in RandomForest for PyMVA.
In the same class MethodPyRandomForest we must support regression, two class classification, multiclass-classification and variable ranking.
You can see how @stwunsch give support for multiclass-classification in AdaBoost by example
a35fdf6
and may you want to rewrite you code according to TMVA design.

Cheers
Omar.

[omazapa]

Hi,
The code dont have the required to be merged, I leave the comments for the @satyarth934
Cheers,
Omar.