[ENH] estimators failing check_regressor_against_expected_results

[TonyBagnall]

Describe the feature or idea you want to propose

no idea on these

• "RDSTRegressor"
• "RISTRegressor"

Describe your proposed solution

fix or create issue

Describe alternatives you've considered, if relevant

No response

Additional context

No response

[ricor07]

Hello, I'd like to join the project by working on this issue.

[baraline]

Hi @ricor07, just a heads-up, this issue is a bit tricky, so if you are new to python/numba/git CI, this might not be the easiest place to start with.

The problem we're facing is that these two estimators (RDST and RIST) are failing a test (check_regressor_against_expected_results, see this file and this one to reproduce the results) on some of the platforms we are testing on (Ubuntu/Windows/MacOS).

The problem is that the expected results can differ (despite fixing a random seed) from platform to platform. The goal of the issue is to understand and fix this, but we didn't have much success on that for now. You can go in the test_config file to make these test run on the CI, otherwise there are excluded.