A feature vector is an ordered collection of measurements used to represent an observation for statistical analysis or pattern recognition. It is commonly written
 |
as a point in a 
-dimensional
feature space, where each component 
is a feature such as a measured quantity, category, or derived
descriptor. Feature vectors make observations suitable as inputs to procedures such
as classification, cluster
analysis, and regression.
A feature vector may contain measurements taken directly from the observation or values produced by feature extraction. For example, an image can be represented by its raw pixel values or by a shorter vector of edges, textures, or coordinates obtained by principal component analysis.
