hops-transformations

To apply data preprocessing transformations in Hopsworks, you can use this Skill to manage built-in functions and Python UDFs for scaling, encoding, and imputation within your machine learning workflows.

Yes, you can use custom Python UDFs for feature engineering in Hopsworks. This Skill enables you to create and apply custom transformations to handle specific data preprocessing needs in your feature views.

Model-dependent transforms in Hopsworks apply learned transformations like scalers and encoders from training data, while on-demand transforms compute feature group transformations dynamically at request time.

You attach a transformation to a feature view in Hopsworks by defining a custom Python UDF or selecting a built-in function, then applying it directly to the feature view for your ML pipeline.

Yes, you need Python knowledge and familiarity with Hopsworks features to effectively manage transformations. This Skill requires understanding Python UDFs and Hopsworks data preprocessing workflows to operate correctly.

Available built-in transformations for machine learning data preprocessing in Hopsworks include scalers, encoders, and imputers, which are applied as model-dependent transforms that learn from your training data.