Labels, features, and train/test splits — step 2 of 7

You're building a churn model from 12 months of user history. Cursor writes train_test_split(X, y, test_size=0.2, random_state=42). What's the problem?