Adam Li

Thanks Thomas! Adding some comments to augment the discussion. The high level goal of all 3 I would say is to enable more flexibility wrt Python/Cython code to allow maximal...

In order to move #22754 along, and possibly get it into a state that can be easily refactored for a stand-alone scikit-learn-contrib package, just wanted to see if there was...

Hey @jjerphan thanks for the reply! I think for me I'm faced with a few obstacles and I'm hoping for some clarity here. Happy to discuss at OH whenever they...

Hi @jjerphan sorry about logging in late. Today was Labor Day in the USA, so I was actually traveling during this time. I tried joining a bit after the meeting...

Linking convo as reference: https://github.com/neurodata/scikit-learn/pull/16 and specifically https://github.com/neurodata/scikit-learn/pull/16#issuecomment-1067516758. I am going to proceed by adding some lightweight unit tests to determine integrability, and getting a working version that we can...

### Progress as of 3/16/22 Added Oblique trees/forests to the existing unit tests parametrizations and they mostly work, with one minor issue on pickling. The following is a summary: -...

I'm investigating the pickling issue. As a reminder, the iris dataset messes up somehow. These are details of my exploration so far: - sample numbers of the iris dataset the...

Crazy weird error on the first sample of the difference (i.e. sample 38 of iris): There is an incorrect addition(?) somehow... est before pickling: ``` decision node 0 : (X_test[0,...

~Ah as it turns out it was a Cython issue, where I was not initializing the value of `j` to 0 when doing `for j in range(...)`. Weird that it...

Okay the error was a slight issue when trying to sample an oblique projection. Fixed in https://github.com/scikit-learn/scikit-learn/pull/22754/commits/a44f1b8b2140790ec7dc9f725676e68a713dbecc