Articles • orbital

Introduction to orbital: Start here if this is your first time using orbital! You will learn how to convert a fitted workflow into an orbital object for lightweight prediction.

Supported Models and recipes steps: A complete list of supported parsnip models, recipes steps, and tailor adjustments that can be converted to orbital objects.
Pros and Cons: Weigh the benefits and limitations of using orbital for prediction to determine if it’s right for your use case.

Supported backends: Examples of running predictions in some of the known supported backends including SQLite, Spark, DuckDB, and Arrow.
Database deployment: Deploy orbital predictions to a database by creating tables or views, with guidance on when to use each approach.
SQL expression sizes: Understand how model type and hyperparameters affect the size of generated SQL, and learn strategies for keeping queries manageable.
Parallel tree evaluation in databases: Use the separate_trees argument to enable parallel evaluation of tree ensembles and avoid expression depth limits in databases.
Float precision at split boundaries: How 32-bit vs 64-bit float differences can affect tree model predictions.