Skip to main content

What is H2O eScorer?

H2O eScorer is a production-ready service for model deployment and scoring that comes with a bundle of feature-rich and intuitive self-service tools to rapidly operationalize your H2O.ai models. It lets you score H2O-3 and Driverless AI MOJOs and Python models concurrently. H2O eScorer is a base component for integration with Snowflake, AWS, GBQ, Qlik, Tableau, Pega and UiPath among others.

H2O eScorer also provides a light-weight, low-latency Python client with simple APIs that your existing enterprise applications can use to make predictions in real time or as a batch.

H2O eScorer offers the following as its main features:

  • Supports Keycloak authentication and user roles.
  • Provides telemetry data on scoring and performance per request or row.
  • Dynamically scalable and fully redundant in the H2O Managed Cloud (HAMC).
  • Supports several different calling formats (for example, URL, JSON, and file).
  • Calls can return predictions and or explainability in a single call.
  • Database scoring is supported directly from H2O eScorer against a database.
  • A GUI option is available that provides access to all of the functions of H2O eScorer.
  • Autogen enables deployment artifacts to be generated for a model so that other applications or user personas can use the models.
  • Support for structured and unstructured models (i.e., tabular and images).

For more detailed information on H2O eScorer, download the H2O eScorer - Using Models in Production PDF.

What's included?​

  1. eScorer: A secure, auto-scalable, and highly available H2O AI Cloud service.
  2. Wave app: A graphical user interface for model registry, model statistics, and lifecycle management.
  3. Python client: We believe that "the best code is no code at all", that is why we offer APIs that are simple and robust. These APIs allow you to make predictions on millions of rows of data with less than 10 lines of code.
info

H2O eScorer is not a core component of H2O AI Cloud and can be installed on a per-invitation basis.

What can I do with it?​

You can use H2O eScorer to do the following:

  • Deploy new models using the point-and-click GUI.
  • View a list of models and model features, and score sample data all within your browser.
  • Automatically generate deployment code for your models, including Snowflake Functions and Jupyter Notebooks, to help you get your models to production quickly.
  • Integrate the Python API with just a few lines of code to score models in your existing enterprise applications.
  • View a real-time dashboard that displays live information on model invocations, latency, errors, and more.
Feedback