Fastest way to set up H2O.ai cluster using Clouderizer Workspace

Original article can be found here (source): Artificial Intelligence on Medium

Fastest way to set up H2O.ai cluster using Clouderizer Workspace

H2O.ai is one of the most popular AutoML platforms, helping citizen data scientists create predictive ML models in minutes. Being free and open source makes it all the more attractive for the growing ML community.

H2O.ai offers Flow UI which is a notebook like environment allowing users to easily import their data, explore and analyse data and generate predictive models from them. Interface is automated and intuitive to ensure users, without advanced data science expertise, can make good progress in creating ML models

Today we will try to set up a single node H2O.ai cluster with Flow UI using Clouderizer Workspace.

Prerequisites

  • Active Clouderizer account (signup here if needed)

Setup

  • Login to your Clouderizer account and go to the Workspace tab.
  • Press “New Project” to create a new Workspace project. Give your project a name, “H2OFlowUI”.
  • In case you wish to pull code from some git repository, enter git URL under Project Location. Leave empty if not needed. Press Next.
  • In case you wish to upload some dataset, enter dataset URL. Leave empty if not needed. Press Next.
  • On next tab, select “Remote Access” tab and make sure to check “H2O.ai (Enable secure webaccess of H2O.ai v3 from Clouderizer console.)”
  • Press Next and Finish.
  • Your Clouderizer Workspace project is now ready. You can run it on your AWS / GCP cloud accounts or locally on your Ubuntu or Mac machines.
  • From the main project list, press the Start button for your project card. This will offer you following choices
  • In case you have linked your AWS or GCP account with Clouderizer, you can select AWS or GCP and run your project there.
  • You can also run this project locally on your Ubuntu or Mac machine. Select macOS from the menu above. This will show you a bash command for starting the project locally.
  • Note: Docker is a prerequisite for running Clouderizer Workspace projects locally on Ubuntu or Mac.
  • Once you run this command in a terminal, project status on Clouderizer console changes to show startup progress. It might take a couple of minutes before the project is in Running state.
  • Once a project is in “Running” state, your H2O v3 single node cluster is all set up. A few buttons like SSH, Jupyter and H2O.ai become active on the project card.
  • Press H2O.ai button on the project card. This should launch H2O Flow UI in a new tab.

Here is a quick article about using H2O Flow UI to develop an ML model and deploying it using Clouderizer Showcase.