Implementation:Scikit learn Scikit learn RandomSampling

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

Overview

Concrete tool for providing random sampling utilities including sampling without replacement and sparse random matrix generation, provided by scikit-learn.

Description

The sklearn.utils.random module provides utilities for random sampling. It re-exports sample_without_replacement from the Cython-optimized _random module and implements _random_choice_csc, which generates sparse random matrices given column class distributions. The sparse matrix generation is useful for creating random multi-label classification targets.

Usage

Use these utilities when you need efficient random sampling without replacement or need to generate sparse random matrices with specific class distributions, particularly in the context of multi-label classification data generation.

Code Reference

Source Location

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

Signature

def _random_choice_csc(n_samples, classes, class_probability=None, random_state=None):

# Re-exported from sklearn.utils._random:
sample_without_replacement

Import

from sklearn.utils.random import sample_without_replacement

I/O Contract

Inputs

Name	Type	Required	Description
n_samples	int	Yes	Number of samples to draw in each column
classes	list of arrays	Yes	List of classes for each column (output)
class_probability	list of arrays or None	No	Class distribution for each column (default None uses uniform)
random_state	int, RandomState or None	No	Controls randomness of sampled classes (default None)

Outputs

Name	Type	Description
random_matrix	sparse csc matrix of shape (n_samples, n_outputs)	Sparse random matrix with sampled class labels

Usage Examples

Basic Usage

from sklearn.utils.random import sample_without_replacement
import numpy as np

# Sample 5 unique indices from a population of 100
indices = sample_without_replacement(100, 5, random_state=42)
print(indices)

Related Pages

• Principle:Scikit_learn_Scikit_learn_Cross_Validation