[BUG] fix the RandomizedSearchCV backend API mismatch.

[MaazAli8460]

Reference Issues/PRs

Fixes #1013.

What does this implement/fix? Explain your changes.

This PR fixes the RandomizedSearchCV backend API mismatch.

Changes made:

Added backend="loky" to RandomizedSearchCV.init in _tuning.py
Passed backend=backend through to the BaseGridSearch constructor so backend selection works as documented.

Does your contribution introduce a new dependency? If yes, which one?

No.

What should a reviewer concentrate their feedback on?

1. Is backend left out intentionally in RandomizedSearchCV constructor?
2. Is the problem in RandomizedSearchCV documentation?

Did you add any tests for the change?

I did not add tests in the commit, but this script is running without errors.

import pandas as pd
from sklearn.model_selection import KFold

from skpro.model_selection import RandomizedSearchCV
from skpro.regression.dummy import DummyProbaRegressor

# Minimal data
X = pd.DataFrame({"x": [1, 2, 3, 4]})
y = pd.DataFrame({"y": [1.0, 2.0, 3.0, 4.0]})


rscv = RandomizedSearchCV(
    estimator=DummyProbaRegressor(),
    cv=KFold(n_splits=2),
    param_distributions={"strategy": ["empirical"]},
    n_iter=1,
)

print(rscv)
print("Backend:", rscv.backend)
rscv.fit(X, y)
print(rscv.cv_results_)

Output

(.venv) D:\D drive\Projects_GSoC_ESoC\skpro>python x.py
RandomizedSearchCV(cv=KFold(n_splits=2, random_state=None, shuffle=False),
                   estimator=DummyProbaRegressor(), n_iter=1,
                   param_distributions={'strategy': ['empirical']})
Backend: loky
   mean_test_CRPS  mean_fit_time                     params  rank_test_CRPS
0            1.75       0.006672  {'strategy': 'empirical'}             1.0

PR checklist

For all contributions

• I've added myself to the list of contributors with any new badges I've earned :-)
  How to: add yourself to the all-contributors file in the skpro root directory (not the CONTRIBUTORS.md). Common badges: code - fixing a bug, or adding code logic. doc - writing or improving documentation or docstrings. bug - reporting or diagnosing a bug (get this plus code if you also fixed the bug in the PR).maintenance - CI, test framework, release.
  See here for full badge reference
• The PR title starts with either [ENH], [MNT], [DOC], or [BUG]. [BUG] - bugfix, [MNT] - CI, test framework, [ENH] - adding or improving code, [DOC] - writing or improving documentation or docstrings.

For new estimators

• I've added the estimator to the API reference - in docs/source/api_reference/taskname.rst, follow the pattern.
• I've added one or more illustrative usage examples to the docstring, in a pydocstyle compliant Examples section.
• If the estimator relies on a soft dependency, I've set the python_dependencies tag and ensured
  dependency isolation, see the estimator dependencies guide.

[Copilot]

Pull request overview

Fixes the documented backend parameter mismatch in RandomizedSearchCV so callers can select a parallel backend as intended.

Changes:

• Exposes backend in RandomizedSearchCV.__init__ and forwards it to BaseGridSearch.
• Sets a default scoring metric (CRPS) when scoring=None to avoid runtime failures.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

scoring is documented as accepting str/callable, but _fit assumes a metric object with .name and .get_tag() (used for scoring_name and ranking). Passing a callable or string will raise at runtime. Consider coercing/validating self.scoring to a BaseMetric (or raising a clear ValueError if unsupported) so the implementation matches the documented API.

Suggested change

	scoring = CRPS()
	scoring = CRPS()
	# validate scoring: BaseGridSearch requires a metric-like object
	# with at least a ``name`` attribute and ``get_tag`` method for
	# result naming and ranking. Passing a bare string or callable will
	# otherwise fail at runtime.
	if not (hasattr(scoring, "name") and hasattr(scoring, "get_tag")):
	raise ValueError(
	f"'scoring' must be a metric-like object with 'name' and "
	f"'get_tag' attributes when used with {self.__class__.__name__}. "
	"Passing strings or generic callables is not supported "
	"in this tuner; please pass a metric object instead."
	)

[Copilot]

Please add a small regression test that instantiates RandomizedSearchCV(..., backend="threading"/"loky") and asserts the attribute is set and used (e.g., by checking no error and that backend is forwarded into parallelize via a simple fit). This prevents future regressions of the documented backend parameter support.

[fkiraly]

Thanks - to prevent regression, could you add a parameter set to get_test_params that sets backend explicitly to be the "none" backend (to avoid nested parallelization)?