Implementation:Scikit learn Scikit learn FunctionTransformer

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Data Preprocessing, Feature Engineering
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for constructing a transformer from an arbitrary callable provided by scikit-learn.

Description

FunctionTransformer forwards its X (and optionally y) arguments to a user-defined function or function object and returns the result. This is useful for stateless transformations such as taking the log of frequencies, doing custom scaling, or any other custom feature transformation. It integrates seamlessly into scikit-learn pipelines.

Usage

Use FunctionTransformer when you need to apply a custom stateless transformation as part of a scikit-learn pipeline. It wraps any callable into a transformer-compatible interface, enabling custom preprocessing steps alongside standard scikit-learn transformers.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/preprocessing/_function_transformer.py

Signature

class FunctionTransformer(TransformerMixin, BaseEstimator):
    def __init__(
        self,
        func=None,
        inverse_func=None,
        *,
        validate=False,
        accept_sparse=False,
        check_inverse=True,
        feature_names_out=None,
        kw_args=None,
        inv_kw_args=None,
    ):

Import

from sklearn.preprocessing import FunctionTransformer

I/O Contract

Inputs

Name	Type	Required	Description
func	callable	No	The callable to use for the transformation. If None, uses the identity function.
inverse_func	callable	No	The callable to use for the inverse transformation. If None, uses the identity function.
validate	bool	No	Whether to validate the input X array before calling func. Default is False.
accept_sparse	bool	No	Whether func accepts a sparse matrix as input. Default is False.
check_inverse	bool	No	Whether to check that func followed by inverse_func leads to the original inputs. Default is True.
feature_names_out	callable or str	No	Determines the feature names out of the transformer. If 'one-to-one', output names equal input names.
kw_args	dict	No	Dictionary of additional keyword arguments to pass to func.
inv_kw_args	dict	No	Dictionary of additional keyword arguments to pass to inverse_func.

Outputs

Name	Type	Description
X_transformed	array-like	The result of applying func to X.

Usage Examples

Basic Usage

from sklearn.preprocessing import FunctionTransformer
import numpy as np

transformer = FunctionTransformer(np.log1p, np.expm1)
X = np.array([[0, 1], [2, 3]])
X_transformed = transformer.fit_transform(X)
print(X_transformed)
# Inverse transform
X_original = transformer.inverse_transform(X_transformed)
print(X_original)

Related Pages

• Principle:Scikit_learn_Scikit_learn_Feature_Transformation