AI Glossary Term
What does “Feature” mean?
Definition
An input variable used by a model to make predictions.
How Feature works
Example
In a model that predicts house prices, the features might be square footage, number of bedrooms, and neighborhood. The model learns how each feature relates to the sale price.
Why it matters
The choice of features strongly shapes what a model can learn. Including features that act as proxies for protected traits can also introduce bias.
Common questions
Is a feature the same as a column?
In table data, yes: each column used as model input is a feature.
Do deep learning models need features?
They still use inputs, but they can learn useful internal features from raw data such as pixels instead of relying on hand-built ones.
How many features should I use?
Enough to capture the signal without adding noise. Too many irrelevant ones can encourage overfitting.