H2O Tabular AI Services, From Data to Deployment
- Predictive AI
- H2O Driverless AI
- H2O Feature Store
- H2O MLOps
This lesson walks through H2O's tabular AI services end to end, starting in Driverless AI where data is pulled from sources like S3, Snowflake, and the H2O Feature Store. It covers automatic visualization, the Predict wizard for building a customer churn model, and interpretation tools like SHAP values, partial dependence plots, and disparate impact analysis for checking fairness. The lesson closes with the Python client, showing how to register the experiment as a model in H2O MLOps, deploy it, and send a live prediction request to the endpoint.
Key takeaways
- How to connect Driverless AI to data sources like S3, Snowflake, and the H2O Feature Store
- What the Predict wizard does to guide less experienced users through building a churn model
- How SHAP values and partial dependence plots explain which features drive a model's predictions
- Why disparate impact analysis matters for checking model fairness across categories like gender or race
- How to register a Driverless AI experiment as a model and deploy it through H2O MLOps
Start owning your AI