take

Add input validation (non-negative y) to RandomForestRegressor.

Isn't that taken care of by the DecisionTreeRegressor instances?

Hey guys! I'm new to open source contributions, and I'd love to contribute to this issue (the reason I took this up as soon as it was posted :)). I'm an avid machine learning enthusiast and scikit-learn user! Any tips for a newcomer like me? Since I'm not fully familiar with the entire codebase, it's taking me a little while to navigate my way around here.

Add input validation (non-negative y) to RandomForestRegressor.

Isn't that taken care of by the DecisionTreeRegressor instances?

import numpy as np
from sklearn.ensemble import RandomForestRegressor

X = np.array([[1, 2, 3]]).T
y = [-1, 0, 1]
rf = RandomForestRegressor(criterion="poisson", random_state=4)
rf.fit(X, y)
rf.predict(X)

gives array([-0.07666667, -0.07666667, -0.07666667]) but should raise ValueError.

@pk1130 Welcome. I hope your motivation keeps on going.
You'll find general advice in the developer's guide for contributing.
The class RandomForestRegressor lives in sklearn/ensemble/_forest.py. Input validation should happen in def fit(..) which is inherited from class BaseForest. So you need to adapt the code there.
I would start with adding the poisson criterion to the docstring of RandomForestRegressor.
Let me know if you need more guidance.

@lorentzenchr Thanks for the welcome and the quick direction to parts of the codebase. I was reading through class RandomForestRegressor just right now. I'll start off with the docstring as you suggested.

@pk1130 Could you open a PR or can I help you somehow?
I'm asking because there is a sprint this Saturday and this issue might be a good candidate for it. I do not intend to take this issue away from you, just informing.

@lorentzenchr Sorry for the delayed response. Last couple days have been crazy with me injuring my ankle ligament. Couldn't get to this sooner. I've updated the docstring as you'd suggested and I'm figuring out how to add input validation. I'm guessing it's fairly trivial but since I'm relatively new to the codebase, just figuring it out. Haven't looked at the test suite yet to expand on it. Let me know if there's some initial work I can get done before today's sprint meeting.

@pk1130 Easy, there is no rush and good recovery for you foot. If you open a PR, it is easier to help out. In order to open one, you push your branch to your forked repository on github, then you can add a pull request to scikit-learn from github (on your fork).

hi @lorentzenchr can I start working on this?

@designer1234 This issue is currently taken by @pk1130 as you can see under Assignees. So you can ask him and offer our help.

Hey @lorentzenchr! Apologies for the delay in getting around to this. Opened a pull request like you'd suggested. Done everything other than expanding the test suite. Let me know what you think and whether you'd be able to review. Thanks!

hi @lorentzenchr @pk1130 ! I'm new to the scikit-learn repo, and would love to help contribute. I was wondering if this ticket was open for additional help? If so, what could I do to help?

@bsun94 Yes, help is welcome. You can either continue #19464 and incorporate the review comments or start fresh. The issue description at the top lists 3 tasks.

thank you @lorentzenchr ! I might just need a bit of time to get more familiar with the code base, but will be putting in some time this & next week.

@bsun94 Great. It is usually advantageous to open/have a PR early such that we can help you there.

Hi there! I'd put in a PR inclusive of a test - please let me know if it looks good!

Closed in #19836.