Reference · Glossary
Feature engineering
Choosing and transforming **input columns** so a model can learn — scaling, encoding categories, or combining fields.
When to use
Tabular ML before deep learning — good features often beat a fancier algorithm on small data.
When not to
When end-to-end deep nets learn representations from raw pixels or text — less manual feature work.
Example
Convert “$1,200” → numeric 1200, add “days_since_signup,” drop leaky columns that peek at the label.