[ ] Scaling, encoding, and imputation are all pipeline steps - [ ] No `fit_transform` on the full dataset before splitting