H2O.AI Demos

This is a collection of the different demos and applications built using the H2O platform.

Deep Water Demo

This application can take an image of a cat, dog or mouse and classify them. It uses Lenet, Inception & Resnet models under the hood and compares the outputs from the three models.

Github

Mortgage Default Prediction

This application is built using the Freddie Mac loans dataset. It is designed to be tool for an mortgage analyst or portfolio manager who wants to look the whole mortgage portfolio, understand overall risks and drill into individual mortgages to understand likelihood of default.

Malicious Domain Identification

This is a DGA (Domain generation algorithms) identification application. DGA are often used to generate malicious domains to breach systems and model will use AI to identify newer domains.

Github

Ask Craig

We hacked ~ 14,000 job titles off Craigslist Bay Area listings to build a model that classifies the unstructured text data of a job title to a given job category. We built this using Sparkling Water and Spark + MLlib.

Github

Churn

This is simple scoring page built with Steam’s scoring service to predict customer churn for Telecom company.

Lending

This app predicts whether someone would qualify for a loan and if they do qualify, then it proposes an appropriate interest rate.

Github

Residuals

This is a demo of a new feature to visualize residuals in H2O to help data scientists evaluate their model performance better.