Driverless AI Training Settings Explained
- Predictive AI
- H2O Driverless AI
This lesson walks through the Training Settings on the Driverless AI experiment page, covering the Accuracy, Time, and Interpretability knobs along with the Scorer metric. It explains how raising Accuracy expands feature engineering, how Time controls experiment length and early stopping, and how Interpretability trades off explainability against model creativity. It shows the experiment preview pane updating live as each knob changes, plus how Driverless AI suggests a scorer like AUC or RMSE based on the target column. It demonstrates this by switching the target from a categorical credit score field to a numeric income field.
Key takeaways
- What the Accuracy knob controls, expanding feature engineering and model selection as its value increases
- How the Time knob sets experiment length and when early stopping kicks in if scores stop improving
- Why a lower Interpretability setting allows more creative, less explainable engineered features
- How Driverless AI automatically suggests a scorer like AUC for classification or RMSE for regression
- The experiment preview pane updates live as each training setting knob changes
Start owning your AI