Handling Class Imbalance with SMOTE

Hard

machine-learning Capital One Stripe JPMorgan

You apply SMOTE to your entire dataset before splitting into train and test sets. Your model shows excellent test set performance but performs poorly in production. What went wrong?

Multiple Choice

SMOTE is not effective for class imbalance problems SMOTE should only be applied to the training set after splitting to avoid leaking information into the test set SMOTE created too many synthetic samples, causing the model to overfit The production data has a different feature distribution unrelated to SMOTE

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

Handling Class Imbalance with SMOTE

Multiple Choice

Related problems

Deep-dive guides