New Feature Pull Request

Description

This PR introduces CategoricalEncoder, a single-column transformer that combines OneHotEncoder (for encoding frequent categories into one-hot binary columns) and TargetEncoder (for encoding categories using the target variable y).

Fixes #2065

Checklist

• I have read the contributing guidelines
• I have added tests that verify the feature works
• I have added necessary documentation:
  • Docstrings for new functions/classes
  • API documentation if applicable
  • Example notebook or script demonstrating usage
• I have added an entry to CHANGES.rst describing the feature
• My code follows the code style of this project
• I have checked my code and corrected any misspellings

If the feature inherits from SingleColumnTransformer:

• get_feature_names_out has been implemented

How Has This Been Tested?

• Added unit tests in skrub/tests/test_categorical_encoder.py covering:
  • Transformation on both Pandas and Polars Series.
  • Validation when target y is missing (y=None).
  • Cloning and isolation when custom one_hot_encoder or target_encoder instances are provided.
  • Integration with ApplyToCols.
  • Scikit-learn estimator compatibility (check_is_fitted, clone).
• Ran local test suites (pytest skrub/tests/test_categorical_encoder.py and pytest skrub/tests/test_single_column_transformer.py).
• Verified code style and formatting with ruff check.

Example Usage

import pandas as pd
from skrub import CategoricalEncoder

s = pd.Series(["a", "b", "a", "c", "d", "e", "a", "b", "c", "d"], name="col")
y = pd.Series([1, 0, 1, 0, 1, 0, 1, 0, 1, 0])

enc = CategoricalEncoder(max_categories=3)
res = enc.fit_transform(s, y)
print(res)