Implementation:Scikit learn Scikit learn Bunch

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

Overview

Concrete tool for providing a dictionary-like container that exposes keys as attributes, provided by scikit-learn.

Description

The Bunch class extends Python's dict by enabling values to be accessed either by key (bunch["value_key"]) or by attribute (bunch.value_key). It is used throughout scikit-learn as the return type for dataset loading functions and other utilities. It also supports deprecated keys with warning messages.

Usage

Use this container when you need a dictionary that also supports attribute-style access, typically when returning structured data from functions. It is the standard return type for load_iris(), fetch_california_housing(), and other dataset loaders.

Code Reference

Source Location

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

Signature

class Bunch(dict):
    def __init__(self, **kwargs):
    def __getitem__(self, key):
    def __setattr__(self, key, value):
    def __getattr__(self, key):
    def _set_deprecated(self, value, *, new_key, deprecated_key, warning_message):

Import

from sklearn.utils import Bunch

I/O Contract

Inputs

Name	Type	Required	Description
**kwargs	any	No	Key-value pairs to populate the Bunch

Outputs

Name	Type	Description
bunch	Bunch	Dictionary-like object with attribute-style access

Usage Examples

Basic Usage

from sklearn.utils import Bunch

b = Bunch(a=1, b=2)
print(b['b'])  # 2
print(b.b)     # 2
b.c = 6
print(b['c'])  # 6

Related Pages

• Principle:Scikit_learn_Scikit_learn_Dataset_Loading