Implementation:Scikit learn Scikit learn FrozenEstimator

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

Overview

Concrete tool for wrapping a fitted estimator to prevent re-fitting, provided by scikit-learn.

Description

The FrozenEstimator class wraps a fitted estimator and freezes it so that calling fit on it has no effect. Methods like fit_predict and fit_transform are also disabled, while all other methods are delegated to the original estimator. This is useful when you have a pre-trained model as a step in a pipeline and want pipeline.fit to skip re-fitting that step.

Usage

Use this when you have a pre-trained or pre-fitted estimator that should be used as-is within a pipeline, without being re-fitted during pipeline training.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/frozen/_frozen.py

Signature

class FrozenEstimator(BaseEstimator):
    def __init__(self, estimator):

Import

from sklearn.frozen import FrozenEstimator

I/O Contract

Inputs

Name	Type	Required	Description
estimator	estimator	Yes	The fitted estimator to be kept frozen

Outputs

Name	Type	Description
predictions	ndarray	Predictions delegated to the wrapped estimator's predict method
transformed	ndarray	Transformations delegated to the wrapped estimator's transform method

Usage Examples

Basic Usage

from sklearn.datasets import make_classification
from sklearn.frozen import FrozenEstimator
from sklearn.linear_model import LogisticRegression

X, y = make_classification(random_state=0)
clf = LogisticRegression(random_state=0).fit(X, y)
frozen_clf = FrozenEstimator(clf)
frozen_clf.fit(X, y)  # No-op, does not re-fit
predictions = frozen_clf.predict(X)  # Delegated to clf.predict

Related Pages

• Principle:Scikit_learn_Scikit_learn_Multiclass_Multioutput