Imbalanced class handling (SMOTE, class weights) in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is an imbalanced class problem in machine learning?

It happens when one class has many more examples than another, making the model biased toward the bigger class.

beginner

What does SMOTE stand for and what does it do?

SMOTE means Synthetic Minority Over-sampling Technique. It creates new synthetic examples for the smaller class to balance the data.

beginner

How do class weights help with imbalanced classes?

Class weights tell the model to pay more attention to the smaller class by making mistakes on it costlier during training.

intermediate

When should you prefer SMOTE over class weights?

Use SMOTE when you want to increase data size by adding synthetic samples. Use class weights when you want to keep data as is but adjust training focus.

intermediate

What is a potential risk of using SMOTE?

SMOTE can create noisy or unrealistic samples if the minority class is very small or complex, which may confuse the model.

What problem does SMOTE solve?

A Too few examples in the minority class

B Too many features in the dataset

C Overfitting on training data

D Missing values in data

SMOTE creates synthetic samples to increase the number of minority class examples.

How do class weights affect model training?

A They increase the learning rate

B They balance the dataset by adding samples

C They reduce the number of features

D They make errors on minority class more costly

Class weights assign higher penalty to mistakes on minority class to balance learning.

Which method adds new data points to balance classes?

A SMOTE

B Class weights

C Feature scaling

D Cross-validation

SMOTE generates synthetic samples to increase minority class size.

What is a downside of using SMOTE?

A It reduces model accuracy

B It removes minority class samples

C It can create unrealistic samples

D It ignores the majority class

SMOTE may create noisy or unrealistic synthetic data if not used carefully.

When might class weights be preferred over SMOTE?

A When you want to add synthetic samples

B When you want to keep original data unchanged

C When dataset is perfectly balanced

D When you want to reduce training time

Class weights adjust training focus without changing the data.

Explain how SMOTE works and why it helps with imbalanced classes.

Think about how adding new examples can help the model see more minority class data.

Describe how class weights influence model training on imbalanced data.

Consider how the model treats mistakes differently for each class.