Feature derivation in a feature store is the process of creating new features from existing ones to improve the performance of a machine learning model.

To derive new features from existing ones in Databricks, first, create a feature store table with the dataset. Then, use the feature store API to retrieve the existing features and apply transformation techniques such as aggregation, normalization, and feature crossing. Finally, create new feature store tables for the derived features and write them using the feature store API.

Feature derivation in a feature store is not possible as it requires manual feature engineering and selection.

To derive new features from existing ones in Databricks, simply list all the features in the dataset and manually create new features based on their relationships.