Dataiku DSSInstall

Overview

This node integrates with the Dataiku DSS API, enabling users to interact programmatically with various Dataiku DSS resources and operations. Specifically, for the Machine Learning - Lab resource and the Get Partial Dependencies of Trained Model operation, the node retrieves all computed partial dependencies for a specified trained machine learning model within a project.

Partial dependencies are useful in interpreting machine learning models by showing how individual features influence the prediction outcome. This node is beneficial when you want to analyze or visualize feature effects on your trained models directly from Dataiku DSS.

Practical example:
You have trained a model in Dataiku DSS and want to understand the impact of certain features on the model's predictions. Using this node, you can fetch the precomputed partial dependency data for that model and use it downstream in your workflow for reporting or further analysis.

Properties

These properties must be provided to specify exactly which trained model's partial dependencies should be fetched.

Output

The output is a JSON array where each item corresponds to the response from the Dataiku DSS API call for the requested operation.

For the Get Partial Dependencies of Trained Model operation, the json output contains the partial dependencies data computed for the specified trained model. This data typically includes feature names, values, and their corresponding partial dependence values, allowing interpretation of the model behavior.

If the operation involves downloading files (not applicable here), binary data would be returned accordingly, but for this operation, the output is purely JSON.

Dependencies

• Requires an active connection to a Dataiku DSS instance.
• Requires valid API credentials (an API key) for authentication with the Dataiku DSS API.
• The node expects the Dataiku DSS server URL and user API key to be configured in the credentials.
• The Dataiku DSS instance must have the relevant project, analysis, ML task, and trained model available and accessible by the API key.

Troubleshooting

• Missing Credentials Error: If the API credentials are not set or invalid, the node will throw an error indicating missing credentials. Ensure the API key credential is properly configured.
• Required Parameter Missing: The node validates required parameters such as Project Key, Analysis ID, ML Task ID, and Model Full ID. Omitting any of these will cause an error specifying which parameter is missing.
• API Request Failures: Network issues, incorrect URLs, or insufficient permissions may cause API request failures. Check connectivity, verify the API key permissions, and ensure the resource identifiers are correct.
• Parsing Errors: If the API returns unexpected data or errors, the node might fail to parse the response. Review the API response and ensure the Dataiku DSS version supports the requested operation.

Links and References

• Dataiku DSS API Documentation – Official API reference for Dataiku DSS.
• Partial Dependence Plots – Explanation of partial dependence plots in machine learning interpretability.
• Dataiku DSS Machine Learning Lab – Overview of ML Lab features in Dataiku DSS.

Note: This summary is based solely on static code analysis of the node implementation and property definitions without runtime execution.