K-Fold Cross-Validation
Averaging scores across folds.
One Split, One Worry
A single train/test split gives just one score. If that split was lucky or unlucky, your estimate could be misleading.
The Big Idea
Cross-validation reuses your data many times, testing on a different slice each round, then averages the scores for a steadier estimate.
All lessons in this course
- Why You Hold Out a Test Set
- train_test_split Done Right
- K-Fold Cross-Validation
- Stop Data Leakage Before It Starts