In machine learning, a feature (or attribute or variable) is an individual measurable property of the phenomenon being observed. A feature is typically a column in your dataset (age, income, pixel intensity at (x,y), etc.), and each data instance is described by a feature vector (a set of features). Good features capture relevant information that can help the model distinguish between outputs. Feature engineering – the process of selecting, transforming, or creating features – can greatly influence model performance. Features can be categorical, numerical, text (which might be vectorized), etc. In deep learning, features in early layers are learned automatically (like edge detectors in images), which are then combined into higher-level features in later layers. In sum, features are the input signals the model uses to learn patterns.