Implementation:Scikit learn Scikit learn ResponseUtils

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

Overview

Concrete utility module for extracting classifier and regressor response values provided by scikit-learn.

Description

The _response module provides functions to uniformly extract response values from classifiers and regressors. It handles predict_proba, decision_function, and predict outputs for binary, multiclass, and multilabel classification scenarios, ensuring consistent shapes and proper handling of positive label selection.

Usage

Use these utilities when building scoring functions, display objects, or any code that needs to extract prediction values from estimators in a standardized way regardless of the response method used.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/utils/_response.py

Signature

def _process_predict_proba(*, y_pred, target_type, classes, pos_label):
    ...

def _process_decision_function(*, y_pred, target_type, classes, pos_label):
    ...

def _get_response_values_binary(
    estimator, X, *, response_method="auto", pos_label=None
):
    ...

def _get_response_values(
    estimator, X, *, response_method, pos_label=None
):
    ...

Import

from sklearn.utils._response import _get_response_values_binary, _get_response_values

I/O Contract

Inputs

Name	Type	Required	Description
estimator	estimator instance	Yes	A fitted classifier or regressor
X	array-like	Yes	Input data to get predictions for
response_method	str or list of str	Yes	The response method(s) to try (e.g., "predict_proba", "decision_function")
pos_label	int, float, bool, or str	No	The positive class label for binary classification

Outputs

Name	Type	Description
y_pred	ndarray	Response values in a consistent format
pos_label	object	The resolved positive label (for binary classification)

Usage Examples

Basic Usage

from sklearn.linear_model import LogisticRegression
from sklearn.datasets import make_classification
from sklearn.utils._response import _get_response_values_binary

X, y = make_classification(random_state=42)
clf = LogisticRegression().fit(X, y)
y_pred, pos_label = _get_response_values_binary(clf, X, response_method="predict_proba")
print(y_pred.shape)  # (100,) - probabilities for positive class

Related Pages

• Principle:Scikit_learn_Scikit_learn_Metric_Evaluation