Transform: Missing & Duplicate Data

Many machine learning algorithms can't handle missing values, so a pipeline must decide what to do with them:

• Delete — drop rows (dropna) or whole columns that are mostly empty. Simple, but you lose information. • Impute — fill the gaps with a sensible estimate: a mean, median, mode, or neighboring value.

The right choice depends on how much is missing and why. Always count first: df.isnull().sum().