Implementation:Scikit learn Scikit learn EstimatorChecks

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Machine Learning, API Compliance Testing
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete utility module for verifying estimator compatibility with the scikit-learn API provided by scikit-learn.

Description

The estimator_checks module provides a comprehensive suite of checks to verify that custom estimators conform to the scikit-learn API contract. It tests cloning, pickling, parameter validation, fit/predict behavior, sparse input handling, multioutput support, and many more aspects. The main entry points are check_estimator and parametrize_with_checks for use with pytest.

Usage

Use check_estimator or parametrize_with_checks when developing custom scikit-learn estimators to ensure they comply with the scikit-learn API. This is the standard way to validate third-party estimators.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/utils/estimator_checks.py

Signature

def check_estimator(estimator=None, generate_only=False, *, on_skip="warn", on_fail="raise",
                    callback=None, expected_failed_checks=None, array_api_dispatch=False):
    ...

def parametrize_with_checks(estimators, *, expected_failed_checks=None):
    ...

# Individual check functions (examples):
# check_estimators_dtypes, check_fit_score_takes_y, check_estimators_overwrite_params,
# check_classifiers_train, check_regressors_train, check_pipeline_consistency, etc.

Import

from sklearn.utils.estimator_checks import check_estimator, parametrize_with_checks

I/O Contract

Inputs

Name	Type	Required	Description
estimator	estimator instance	Yes	The estimator to validate against the scikit-learn API
generate_only	bool	No	If True, return a generator of (estimator, check) tuples
on_skip	str	No	Action when a check is skipped: "warn" or "pass"
on_fail	str	No	Action on failure: "raise" or "warn"
expected_failed_checks	dict	No	Dictionary of expected failures for known issues

Outputs

Name	Type	Description
None	None	Raises errors or warnings if checks fail
generator	generator	If generate_only=True, yields (estimator, check) tuples

Usage Examples

Basic Usage

from sklearn.utils.estimator_checks import check_estimator
from sklearn.linear_model import LogisticRegression

# Run all API compliance checks
check_estimator(LogisticRegression())

# Use with pytest parametrize
from sklearn.utils.estimator_checks import parametrize_with_checks

@parametrize_with_checks([LogisticRegression()])
def test_sklearn_compatible_estimator(estimator, check):
    check(estimator)

Related Pages

• Principle:Scikit_learn_Scikit_learn_Metric_Evaluation