What is the term for when Vertex AI infers how to use a feature based on its data type and values?

The term that describes when Vertex AI infers how to use a feature based on its data type and values is transformation. In machine learning, transformation encompasses various methods of processing input data to make it suitable for the model. When the system identifies the characteristics of features, it can automatically apply appropriate transformations that enhance the model's ability to learn from the given data.

For example, if a feature is recognized as a categorical variable, Vertex AI may automatically encode it into a format that is more useful for modeling, such as converting it into numerical representations. Similarly, if a feature is numeric, it might determine the best way to scale or log-transform the values based on their distribution.

This automatic adjustment helps streamline the preparation process, ensuring that the model gets the most relevant and properly formatted inputs for learning, thereby enhancing overall performance. Other options like normalization and standardization specifically refer to scaling numeric data to a certain range or distribution but do not capture the broader concept of inferring the best representation of a feature based on its type and values. Feature engineering, while similar, is more about manually creating new features rather than inferring how to use existing ones automatically.