Implementation:Scikit learn Scikit learn Encode

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

Overview

Concrete utility module for encoding categorical values and finding unique elements provided by scikit-learn.

Description

The _encode module provides helper functions for finding unique values in arrays, including support for Python object dtypes and proper NaN handling. It includes _unique, _unique_np, and _unique_python functions that correctly handle missing values and work with both numpy arrays and Array API-compatible backends.

Usage

Use these utilities when you need to find unique values in label or feature arrays, particularly when NaN values or object dtypes are present, such as during label encoding or ordinal encoding operations.

Code Reference

Source Location

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

Signature

def _unique(values, *, return_inverse=False, return_counts=False):
    ...

def _unique_np(values, return_inverse=False, return_counts=False):
    ...

def _unique_python(values, return_inverse=False, return_counts=False):
    ...

Import

from sklearn.utils._encode import _unique

I/O Contract

Inputs

Name	Type	Required	Description
values	ndarray	Yes	Values to check for unique elements
return_inverse	bool	No	If True, also return the indices of the unique values
return_counts	bool	No	If True, also return the number of times each unique item appears

Outputs

Name	Type	Description
unique	ndarray	The sorted unique values
unique_inverse	ndarray	Indices to reconstruct the original array from unique (if requested)
unique_counts	ndarray	Number of times each unique value appears (if requested)

Usage Examples

Basic Usage

import numpy as np
from sklearn.utils._encode import _unique

values = np.array([3, 1, 2, 3, 1, 2])
uniques = _unique(values)
print(uniques)  # array([1, 2, 3])

uniques, inverse = _unique(values, return_inverse=True)
print(inverse)  # array([2, 0, 1, 2, 0, 1])

Related Pages

• Principle:Scikit_learn_Scikit_learn_Feature_Transformation