Adam Li

> * On top of the API changelog entry, we might want to also include a `versionchanged` directive for each of the attributes on those estimators. The `tree_.value` is currently...

The issues raised are fixed now: 1. only show floating point up to 3 decimal places when printing `tree_.value` in the example 2. show proportions when plotting tree (here I...

Kay that makes sense! Done

Thanks for the review @lesteve! I have addressed your comments in [7343096](https://github.com/scikit-learn/scikit-learn/pull/29331/commits/7343096c0d6dacbd0cc347ca882881f0e63bc41f)

Will be closing this in favor of #25448 that branches off from the refactoring changes in #24678 and #24990 so the PR demonstrates the combination of all three refactoring.

@jjerphan as a quick proof of concept, I'm pushing up a "pseudocode-demo" example of how the refactored Tree API would enable a quantile tree. See: `sklearn/tree/_quantile_tree_demo.pyx`. My impression is that...

> Do you need some guidance for this contribution? Hey @jjerphan! Sorry been busy and I have a deadline coming up so haven't had the chance to work on this...

Hey @Micky774 yeah I'm interested in helping out with this push, but then got a bunch out on my plate. Once neurips rebuttal period is over I was going to...

> I don't know if this YAGNI or not. By definition, having `max_feature > 1` and `bootstrap=True` would not be a bootstrap by definition anymore. Is there any concrete example...

To add to the motivation and use-cases described above: The typical training paradigm of 80/20 split used in machine learning is pretty sensible. The train/test split used in cross-validation can...