Explaining your AutoML model

Knowing your results is important, but knowing how your model derived its results is just as integral to working with machine learning. Here is where model explainability plays a key role. Explainability is the ability to say which features are most important in building your AutoML model. This is especially important in industries where you have to be able to legally explain your machine learning models, for example, if you built a model to determine who is approved for a loan:

1. To begin, click the Explanations tab next to Metrics.
2. Click the first ID under Explanation ID on the right-hand side of the screen.
3. Click the slider button next to View previous dashboard experience.
4. Click Global Importance.

   Immediately, you will see your columns ranked in order of importance. Sex is the most important column, followed by Pclass and Age, as shown in Figure 3.17. With an importance value of 1.1, Sex is roughly twice as important as Pclass, with...