Loading…
Glossary term
A measurable property of the data that a model uses as input.
A feature is one measurable property used as input, such as a house's area, a patient's age or a pixel's brightness. A model sees each example as a list of feature values.
Deep learning can learn its own features from raw data, while classic methods rely on features chosen by people.
| Area (m²) | Bedrooms | Age (yrs) | |
|---|---|---|---|
| House A | 120 | 3 | 15 |
| House B | 85 | 2 | 40 |
| House C | 210 | 5 | 5 |
Good features make learning easier. Choosing the wrong ones, or leaking the answer into them, is a common cause of misleading results.