Skip to main content
All terms
Data

Feature Engineering

Crafting and selecting the input variables that help a model learn better.

Definition

Feature engineering is the work of creating, selecting, and shaping the input variables (features) a model learns from — for example, turning a raw date into 'day of week' or combining two columns. Good features often improve results more than a fancier model. Deep learning reduces but does not eliminate the need for it.