A synthetic feature formed by multiplying two or more features is called what?

A synthetic feature formed by multiplying two or more features is referred to as a Feature Cross. This term describes a specific technique used in feature engineering to create new features that can capture interactions between existing features. By multiplying features together, you enable a model to learn more complex relationships within the data. This is particularly useful in cases where the interaction between variables is important for prediction, as it allows the model to consider how these features jointly influence the target variable.

Feature crossing can significantly enhance the predictive power of models, especially in scenarios like linear models or decision trees, where interactions may not be inherently recognized. Utilizing feature crosses can lead to improved performance, making it a powerful approach in machine learning workflows where capturing complexities in the data is crucial.

The other choices—Composite Feature, Categorical Feature, and Derived Feature—do not specifically describe the process of multiplying features to create a new synthetic feature. A Composite Feature typically refers to a combination of multiple features into a single unit, but it does not specifically denote multiplication. Categorical Features are variables that represent categories or groups, and Derived Features are often used more broadly and can include any type of feature created from existing data, not limited to multiplication.