Switch RandomForest max_depth default to None

[csadorf]

Changes RandomForestClassifier, RandomForestRegressor, and their Dask variants to use max_depth=None as the default now that unlimited-depth support is available. Also removes the 26.06 deprecation sentinel and updates tests/docs to expect the new default.

Validation:

• python -m pytest python/cuml/tests/test_random_forest.py -k "default_parameters or unlimited_max_depth"
• python -m pytest python/cuml/tests/dask/test_dask_random_forest.py -k "unlimited_max_depth"
• python -m pytest python/cuml/tests/test_sklearn_compatibility.py -k "RandomForest"

Closes #7946

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 393dfb5c-2928-4b38-a1be-3ee8379d3780

📥 Commits

Reviewing files that changed from the base of the PR and between cf6a546 and df38839.

📒 Files selected for processing (18)

• python/cuml/cuml/dask/ensemble/base.py
• python/cuml/cuml/dask/ensemble/randomforestclassifier.py
• python/cuml/cuml/dask/ensemble/randomforestregressor.py
• python/cuml/cuml/ensemble/randomforest_common.pyx
• python/cuml/cuml/ensemble/randomforestclassifier.py
• python/cuml/cuml/ensemble/randomforestregressor.py
• python/cuml/tests/dask/test_dask_random_forest.py
• python/cuml/tests/explainer/test_explainer_common.py
• python/cuml/tests/explainer/test_explainer_kernel_shap.py
• python/cuml/tests/explainer/test_gpu_treeshap.py
• python/cuml/tests/test_api.py
• python/cuml/tests/test_base.py
• python/cuml/tests/test_common.py
• python/cuml/tests/test_fil.py
• python/cuml/tests/test_meta_estimators.py
• python/cuml/tests/test_pickle.py
• python/cuml/tests/test_random_forest.py
• python/cuml/tests/test_sklearn_compatibility.py

💤 Files with no reviewable changes (9)

• python/cuml/tests/explainer/test_gpu_treeshap.py
• python/cuml/tests/test_common.py
• python/cuml/tests/dask/test_dask_random_forest.py
• python/cuml/tests/explainer/test_explainer_common.py
• python/cuml/tests/test_fil.py
• python/cuml/cuml/dask/ensemble/base.py
• python/cuml/tests/explainer/test_explainer_kernel_shap.py
• python/cuml/tests/test_meta_estimators.py
• python/cuml/tests/test_pickle.py

---

📝 Walkthrough

Summary by CodeRabbit

• Documentation

  • Updated Random Forest classifier and regressor documentation to reflect new default parameter values and version information.

• Bug Fixes

  • Removed deprecation warnings previously displayed during model fitting.
  • Simplified parameter validation logic in Random Forest models.
  • Updated test suites and warning filters across the codebase to align with revised defaults.

Walkthrough

This PR removes the max_depth=16 deprecation cycle by eliminating the deprecated sentinel value, associated FutureWarning logic, and all test warning suppressions. The default max_depth parameter changes from 16 to None (unlimited depth) across CPU and Dask RandomForest estimators, with updated docstrings and test expectations.

Changes

Default max_depth switch and deprecation removal

Layer / File(s)	Summary
Core CPU RandomForest API default switch python/cuml/cuml/ensemble/randomforestclassifier.py, python/cuml/cuml/ensemble/randomforestregressor.py, python/cuml/cuml/ensemble/randomforest_common.pyx	RandomForestClassifier and RandomForestRegressor constructor parameters change from "deprecated" sentinel to None. BaseRandomForestModel in the cython layer updates its default to None and removes sentinel-to-16 mapping in _params_to_cpu. Docstrings updated to reflect new default and versionchanged:: 26.08 directive.
Dask RandomForest API and deprecation logic removal python/cuml/cuml/dask/ensemble/base.py, python/cuml/cuml/dask/ensemble/randomforestclassifier.py, python/cuml/cuml/dask/ensemble/randomforestregressor.py	Dask BaseRandomForestModel removes max_depth user-tracking logic, FutureWarning emission in _fit, and set_params bookkeeping. Dask classifier and regressor docstrings updated to document None default.
Core test expectations and warning filters python/cuml/tests/test_random_forest.py, python/cuml/tests/test_sklearn_compatibility.py	test_random_forest.py removes global warning filter, removes pytest.warns() expectations around fit calls, and adds assertions verifying max_depth=None in get_params(). test_sklearn_compatibility.py removes explicit max_depth=None from ESTIMATORS instantiation.
Test warning filter updates across suite python/cuml/tests/{dask/test_dask_random_forest.py, test_api.py, test_base.py, test_common.py, test_meta_estimators.py, test_pickle.py, test_fil.py, explainer/test_*.py}	Remove module-level pytestmark filters for max_depth deprecation FutureWarning across all affected test modules; replace with alternative warning filters where needed (e.g., as_fil() deprecation in test_fil.py, forest-inference in test_explainer_common.py).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• rapidsai/cuml#7895: Implements max_depth=None support for unlimited depth in RandomForest estimators; this PR completes the deprecation cycle by removing the deprecated sentinel and switching the default.
• rapidsai/cuml#7958: Related deprecation removal PR that handles the same max_depth deprecation plumbing across CPU and Dask variants.
• rapidsai/cuml#8113: Adds rapids-pre-commit-hooks suppression comments around the max_depth deprecation warning block that this PR removes entirely.

Suggested labels

sklearn-api-compat

Suggested reviewers

• jcrist
• hcho3

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 14.29% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: switching RandomForest max_depth default from 16 to None.
Description check	✅ Passed	The description clearly explains the changes made: default max_depth switch to None, removal of deprecation sentinel, and updates to tests/docs.
Linked Issues check	✅ Passed	All requirements from issue #7946 are met: max_depth default changed to None in all variants [#7946], deprecation warning removed [#7946], and documentation/tests updated [#7946].
Out of Scope Changes check	✅ Passed	All changes are directly related to switching max_depth default to None; no unrelated or out-of-scope modifications detected.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[jcrist]

[jcrist]

/merge