custom-sklearn-estimator

What problem does it solve?

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.

Core Features & Use Cases

• Minimal init with keyword arguments and defaults; assigns each parameter to a corresponding attribute.
• Implement fit(self, X, y=None, **kwargs) and return self, while validating inputs and creating learned attributes with trailing underscores (e.g., coef_, classes_).
• Implement prediction/transform methods that use check_is_fitted and validate inputs with check_array, ensuring compatibility with pipelines and estimator checks.
• Expose parameters via get_params/set_params and support randomness via random_state and check_random_state.
• Prepare for estimator checks and optional tagging through sklearn_tags and compatibility helpers.

Quick Start

Create a minimal sklearn-compatible estimator by defining init, fit, and predict methods following the rolling your own estimator rules.