Exploring the various explainability features
The output of the explainability interface is an H2OExplanation object. The H2OExplanation object is nothing but a simple dictionary with the explainability features’ names as keys. You can retrieve individual explainability features by using a feature’s key name as a dict key on the explainability object.
If you scroll down the output of the explainability interface for the H2O AutoML object, you will notice that there are plenty of headings with explanations. Below these headings, there’s a brief description of what the explainability feature is. Some have graphical diagrams, while others may have tables.
The various explainability features are as follows:
- Leaderboard: This feature is a leaderboard comprising all trained models and their basic metrics ranked from best performing to worst. This feature is computed only if the explainability interface is run on the H2O AutoML object or list of...
