Stratified K-fold in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is Stratified K-fold cross-validation?

Stratified K-fold is a way to split data into K parts (folds) so that each fold has the same proportion of each class as the whole dataset. This helps keep the data balanced during training and testing.

beginner

Why use Stratified K-fold instead of regular K-fold?

Regular K-fold may create folds with uneven class distributions, which can cause biased model evaluation. Stratified K-fold keeps class proportions consistent, giving more reliable results especially for imbalanced data.

intermediate

How does Stratified K-fold handle imbalanced datasets?

It ensures each fold has roughly the same percentage of samples from each class as the full dataset, so minority classes are fairly represented in every fold.

advanced

In Stratified K-fold, what happens if the number of samples in a class is less than the number of folds?

Some folds may not contain samples from that class because there aren't enough samples to distribute evenly. This can affect the stratification quality.

beginner

Write a simple Python code snippet to perform Stratified K-fold cross-validation using scikit-learn.

from sklearn.model_selection import StratifiedKFold
import numpy as np

X = np.array([[1], [2], [3], [4], [5], [6]])
y = np.array([0, 0, 1, 1, 0, 1])

skf = StratifiedKFold(n_splits=3)
for train_index, test_index in skf.split(X, y):
    print("TRAIN:", train_index, "TEST:", test_index)

What is the main goal of Stratified K-fold cross-validation?

A To randomly shuffle data without any class consideration

B To keep class proportions balanced in each fold

C To increase the number of folds beyond the dataset size

D To use only one fold for training

Which type of problem benefits most from Stratified K-fold?

A Classification with imbalanced classes

B Regression with continuous targets

C Unsupervised clustering

D Data with no labels

If you have 5 folds, how does Stratified K-fold split the data?

A Each fold has random samples ignoring class

B Each fold has all samples of one class

C Each fold has roughly 1/5 of each class

D Each fold has only majority class samples

What happens if a class has fewer samples than the number of folds in Stratified K-fold?

A Some folds may not contain samples from that class

B The class samples are duplicated to fill all folds

C The class is ignored during splitting

D The number of folds is automatically reduced

Which Python library provides StratifiedKFold for easy use?

A NumPy

B TensorFlow

C Matplotlib

D scikit-learn

Explain how Stratified K-fold cross-validation works and why it is useful for classification problems.

Describe a situation where using Stratified K-fold is better than regular K-fold and why.