explainers

Recipes for Machine Learning Interpretability in Driverless AI

• What are MLI Custom Recipes?
• Best Practices for MLI Recipes
• Resources
• MLI Recipes

What are MLI Custom Recipes?

Driverless AI provides robust interpretability of machine learning models to explain modeling results in a human-readable format. In the Machine Learning Interpetability (MLI) view, Driverless AI employs a host of different techniques and methodologies for interpreting and explaining the results of its models.

The set of MLI techniques and methodologies can be extended with recipes. With MLI BYOR, you can use your own recipes in combination with or instead of all built-in recipes. This allows you to further extend MLI explainers in addition to out-of-the-box techniques.

Custom explainer recipes can be uploaded into Driverless AI at runtime without having to restart the platform, just like a plugin.

Best Practices for MLI Recipes

For security, safety and performance best practices please refer to developer guide

Resources

• Tutorials:
  • Creating Custom Explainer with MLI BYORs
• Notebooks:
  • MLI BYORs Python client in examples
• Documentation:
  • Developer guide
  • API reference
  • MLI documentation

MLI Recipes

IID explainers:

• Morris sensitivity analysis

Explainer templates:

• Feature importance explainer template
• PD/ICE explainer template
• Decision tree explainer template
• Markdown report explainer template
• Markdown report with Vega charts template
• Markdown feature importance summary explainer template
• Scatter plot explainer template

Explainer examples:

• Hello, World! explainer example
• Filter and score explainer example
• Dataset and EDA explainer example
• Explainer parameters example
• Explainer logging example
• Explainer persistence example
• Explainer compatibility check example
• Custom explanation example
• Driverless AI metadata explainer example