Manage Package Dependencies in Workbench

You can analyze data using Workbench Notebooks with packages. If your analysis requires additional packages, you can manage package dependencies to compute. Workbench notebooks do not support native pip installations from notebook cells.

To manage package dependencies:

1. Confirm that the package is approved for the data and analysis type.
2. Create or update an approved dependency file in the workspace. A file named requirements.txt is a common convention.
3. Attach the dependency file to the approved compute cluster as a workspace-backed library when that pattern is configured.
4. Start or restart the approved compute cluster.
5. Verify the changes by running a small cell or notebook to confirm that the package imports successfully. If the changes are applied, run a full analysis.

See Using Oracle AI Data Platform Workbench Notebooks for more information about using Workbench Notebooks to analyze data.