Understand Variable Importance in Driverless AI

This lesson walks through the Variable Importance view in H2O Driverless AI, showing how automatic feature engineering creates new columns through transformers for numeric, categorical, time, text, and image data. It explains that variables are ranked by how much they contribute to the prediction, and that the full ranked list is available in the downloaded summary and logs once an experiment finishes. It covers the diagnostic charts Driverless AI generates depending on the problem type, including ROC, precision recall, lift, and gains charts for classification, and residuals plus actual versus predicted charts for regression.