Why You Should Consider H2O.ai for Your ML Stack

When we talk about machine learning tools, the triad of Tensor Flow, Scikit-Learn, and Spark MLLib. immediately comes to mind.

However, there’s a tool that has been quietly gaining ground: H2O.ai.

This tool was originally born in 2011 with the primary objective for its team to democratize and scale machine learning through a more visual platform that would offer a good user experience, regardless of their technical level.

Some features of H2O.ai:

• Open-Source Java Platform
• Many on-the-shelf models like regression models, ensembles, Deep Learning, PCA, etc.
• Extensive documentation on the solution
• Tutorials and more usage tutorials
• Native REST API
• Integration with R and Python
• Integration with Spark
• Distributed ML model processing
• Web interface via Flow GUI
• Can be used either as a standalone desktop application or via a cluster
• Active community discussing and improving the solution
• Datasource-agnostic support
• Model deployment via plain-old Java objects (POJO)

Below are some videos of H2O.ai in action:

H2O.ai for fraud detection

http://www.youtube.com/watch?v=RqkheMI3Ciw

Customer Churn using H2O.ai http://www.youtube.com/watch?v=-u–LeFltk4

Main discussion forum

Github Repository

https://github.com/h2oai

In future posts, we will discuss some architectural issues and dive into some tutorials.