Implementation:Scikit learn Scikit learn RFE

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Feature Selection, Model Evaluation
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for feature ranking with recursive feature elimination provided by scikit-learn.

Description

RFE (Recursive Feature Elimination) selects features by recursively considering smaller and smaller sets of features. First, the estimator is trained on the initial set of features and the importance of each feature is obtained. Then, the least important features are pruned from the current set. This process is repeated recursively until the desired number of features is reached. The module also provides RFECV, which performs RFE with cross-validation to automatically select the optimal number of features.

Usage

Use RFE when you want to select features by recursively removing the least important ones based on an external estimator's feature weights. Use RFECV when you additionally want to automatically determine the optimal number of features through cross-validation. Both are effective with models that expose feature importance via coef_ or feature_importances_ attributes.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/feature_selection/_rfe.py

Signature

class RFE(SelectorMixin, MetaEstimatorMixin, BaseEstimator):
    def __init__(
        self,
        estimator,
        *,
        n_features_to_select=None,
        step=1,
        verbose=0,
        importance_getter="auto",
    ):

class RFECV(RFE):
    def __init__(
        self,
        estimator,
        *,
        step=1,
        min_features_to_select=1,
        cv=None,
        scoring=None,
        verbose=0,
        n_jobs=None,
        importance_getter="auto",
    ):

Import

from sklearn.feature_selection import RFE
from sklearn.feature_selection import RFECV

I/O Contract

Inputs

Name	Type	Required	Description
estimator	estimator instance	Yes	A supervised learning estimator with a fit method that provides feature importance via coef_ or feature_importances_.
n_features_to_select	int or float	No	Number of features to select. If None, half of the features are selected. Default is None.
step	int or float	No	Number (or fraction) of features to remove at each iteration. Default is 1.
verbose	int	No	Controls verbosity of output. Default is 0.
importance_getter	str or callable	No	How to get feature importances. Default is "auto".
cv	int, cross-validation generator, or iterable	No	Cross-validation splitting strategy for RFECV. Default is None (5-fold).
scoring	str or callable	No	Scoring strategy for RFECV. Default is None (estimator's default scorer).

Outputs

Name	Type	Description
X_transformed	ndarray	The input data with only the selected features retained.
support_	ndarray of shape (n_features,)	Boolean mask of selected features.
ranking_	ndarray of shape (n_features,)	Feature ranking (1 = selected).
n_features_	int	The number of selected features.
estimator_	estimator instance	The fitted estimator used to select features.

Usage Examples

Basic Usage

from sklearn.feature_selection import RFE
from sklearn.svm import SVC
from sklearn.datasets import make_classification

X, y = make_classification(n_samples=200, n_features=20, n_informative=5, random_state=42)

estimator = SVC(kernel="linear")
selector = RFE(estimator, n_features_to_select=5, step=1)
selector = selector.fit(X, y)
print(f"Selected features: {selector.support_}")
print(f"Feature ranking: {selector.ranking_}")
X_selected = selector.transform(X)
print(f"Shape after selection: {X_selected.shape}")

Related Pages

• Principle:Scikit_learn_Scikit_learn_Feature_Selection