Glossary term
Glossary term
Foundations
A feature with a finite set of possible values. For example, a feature whose values may only be animal, vegetable, or mineral is a discrete (or categorical) feature.
Contrast with continuous feature.
Created for this library
A retail data team uses discrete features like product category and store region in its boosted-tree model for sales forecasting.
A churn team uses discrete features like plan tier and payment method as one-hot encoded inputs to its production model.
A search ranking team uses discrete features like device type and country in its ranker alongside continuous engagement signals.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License