Implementation:Scikit learn Scikit learn DummyClassifier

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

Overview

Concrete tool for baseline dummy classifiers and regressors that make predictions ignoring input features provided by scikit-learn.

Description

DummyClassifier makes predictions that ignore the input features, serving as a simple baseline to compare against more complex classifiers. It supports multiple strategies: 'most_frequent' (always predict the most common class), 'prior' (predict based on class prior distribution), 'stratified' (random predictions matching class distribution), 'uniform' (random uniform predictions), and 'constant' (always predict a specified constant). The module also includes DummyRegressor which provides analogous baseline strategies for regression tasks.

Usage

Use DummyClassifier and DummyRegressor as baselines to verify that your actual models are performing better than simple heuristics. Any model that cannot beat the dummy baseline is not learning useful patterns from the data.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/dummy.py

Signature

class DummyClassifier(MultiOutputMixin, ClassifierMixin, BaseEstimator):
    def __init__(self, *, strategy="prior", random_state=None, constant=None):

class DummyRegressor(MultiOutputMixin, RegressorMixin, BaseEstimator):
    def __init__(self, *, strategy="mean", constant=None, quantile=None):

Import

from sklearn.dummy import DummyClassifier, DummyRegressor

I/O Contract

Inputs (DummyClassifier)

Name	Type	Required	Description
strategy	str	No	Prediction strategy: 'most_frequent', 'prior', 'stratified', 'uniform', 'constant' (default='prior')
random_state	int, RandomState, or None	No	Random state for 'stratified' and 'uniform' strategies
constant	str, int, or array-like	No	Constant value to predict when strategy='constant'

Inputs (DummyRegressor)

Name	Type	Required	Description
strategy	str	No	Prediction strategy: 'mean', 'median', 'quantile', 'constant' (default='mean')
constant	float or array-like	No	Constant value to predict when strategy='constant'
quantile	float	No	Quantile to predict when strategy='quantile' (value in [0.0, 1.0])

Outputs

Name	Type	Description
classes_	ndarray of shape (n_classes,)	Unique class labels (DummyClassifier only)
n_classes_	int or list	Number of classes for each output (DummyClassifier only)
class_prior_	ndarray of shape (n_classes,)	Frequency of each class (DummyClassifier only)
n_outputs_	int	Number of outputs
constant_	ndarray	Mean, median, quantile, or constant value predicted (DummyRegressor only)
n_features_in_	int	Number of features seen during fit

Usage Examples

Basic Usage

from sklearn.dummy import DummyClassifier
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split

X, y = load_iris(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42)

dummy = DummyClassifier(strategy="most_frequent")
dummy.fit(X_train, y_train)
print(f"Baseline accuracy: {dummy.score(X_test, y_test):.3f}")

Related Pages

• Principle:Scikit_learn_Scikit_learn_Metric_Evaluation