custom-sklearn-estimator

name	custom-sklearn-estimator
description	Build scikit-learn compatible custom estimators by following the official “rolling your own estimator” rules for __init__, fit/predict, validation, learned attributes, tags, and estimator checks; prerequisite for autogluon-sklearn-wrapper or any sklearn-facing wrappers.

Custom scikit-learn Estimator

Purpose

Create scikit-learn compatible estimators that work with pipelines, model selection, and validation tooling. This skill codifies the required API patterns for __init__, fit, prediction/transform methods, learned attributes, and estimator checks.

Usage

• “rolling your own estimator”
• “custom scikit-learn estimator”
• “build sklearn-compatible class”

Instructions

1. Choose the estimator type and mixins
   • Use ClassifierMixin, RegressorMixin, TransformerMixin, or ClusterMixin as needed, with BaseEstimator last in the inheritance list.
   • For meta-estimators, ensure sub-estimator params are exposed through get_params/set_params (handled by BaseEstimator).
2. Implement a minimal __init__
   • Keyword args with defaults; no validation or logic.
   • Assign each parameter to an attribute with the exact same name.
   • Avoid mutable defaults; do not set attributes with trailing _ here.
3. Implement fit
   • Signature: fit(self, X, y=None, **kwargs) and accept y=None even for unsupervised estimators.
   • Validate inputs using validate_data/check_array; ensure X.shape[0] == y.shape[0] when supervised.
   • Set learned attributes with trailing _ (e.g., coef_, classes_).
   • Return self and overwrite learned attributes on every call unless warm_start=True.
4. Implement prediction/transform methods
   • Call check_is_fitted and validate inputs with validate_data(..., reset=False).
   • Classifiers must use self.classes_ and return labels, not indices.
   • Transformers must preserve sample count and order.
5. Handle randomness correctly
   • Accept random_state=None in __init__, store it unmodified.
   • In fit, use check_random_state and store RNG in random_state_ if needed later.
6. Optional: tags and set_output
   • Implement __sklearn_tags__ if default tags are not appropriate.
   • For transformers, consider get_feature_names_out and set_output compatibility.
7. Validate with estimator checks
   • Run check_estimator or parametrize_with_checks when possible.
   • Use the response checklist in ./templates/estimator-checklist.md to confirm compliance.