Introduced on day 9
ColumnTransformer
A scikit-learn component that applies different preprocessing steps to different column groups and concatenates the results, usually placed inside a Pipeline.
In detail
Typical use: impute and scale numeric columns, one-hot encode low-cardinality categories, target-encode high-cardinality ids. Columns not listed are dropped by default (remainder="drop"). Inside a Pipeline, all its steps are fit only on the rows passed to fit.