Handling Missing Data

Summary

Strategies (drop, impute, flag) for dealing with nulls in a dataset — a decision that materially affects model performance and must be made deliberately, not automatically.

• 1Practice `.isnull()`, `.dropna()` and `.fillna()` with different imputation strategies.
• 2Learn when dropping rows is safe vs. when it introduces bias.
• 3Compare mean/median imputation vs. model-based imputation on a real dataset.