A feature vector is an n-dimensional vector of numerical features that represent some object or instance. In machine learning, input data is often represented as a feature vector – for example, for a house price prediction model, one might create a feature vector like [square_feet, number_of_bedrooms, age_of_house, distance_to_city_center]. In image recognition, after feature extraction, an image might be represented by a vector of learned features (like outputs of a CNN layer). These vectors can be seen as points in an n-dimensional feature space. A dataset is then a collection of feature vectors (usually with corresponding labels for supervised tasks). The term highlights that we’re interested in a numerical encoding of objects that algorithms can operate on. When doing any ML, a key step is converting raw data (which might not be numerical) into feature vectors (via encoding, embedding, etc.) that capture properties of the data relevant to the task.