MLI Custom Recipes

The techniques and methodologies used by Driverless AI for model interpretation can be extended with recipes (Python code snippets). You can use your own recipes in combination with or in place of DAI’s built-in recipes. This lets you extend the capabilities of MLI explainers and out of the box interpretation techniques. The following steps describe how to upload and enable custom recipes in the Machine Learning Interpretability (MLI) view.

참고

For more information on MLI custom recipes including best practices, tutorials, explainer templates, and explainer examples, see the official Recipes for Machine Learning Interpretability in Driverless AI repository.

To upload a custom recipe:

1. Navigate to the MLI page and click the New Interpretation button. Select Upload MLI Recipe from the drop-down menu. You can also select MLI Recipe URL to load a recipe from a raw file, a GitHub repository / tree, or a local directory. Alternatively, you can access these options from the MLI Expert Settings panel.

To enable a custom recipe:

1. On the MLI page, click New Interpretation. Select New Interpretation from the menu.

2. Select a model and a dataset for the interpretation, then click Recipes.

3. Select the custom recipe(s) you want to enable, then click Done.

4. Click the Launch MLI button.