H2O Floats New Release Of Its Predictive Analytics Platform

H2O is shipping a new release of its in-memory prediction engine software that company executives describe as a significant overhaul of the product, including a new modular architecture, new user interface, integration with the Spark processing engine and easier deployment options.

The 3.0 release, according to Mountain View, Calif.-based H2O, takes the product's machine learning capabilities to the next level and helps developers and data scientists build smarter big data applications.

"Prediction is the new search and machine learning is the new SQL," said CEO Srisatish Ambati in an interview.

[Related: 2015 Big Data 100: Business Analytics]

Sponsored post

Businesses use H2O's software to develop predictive analysis applications for such tasks as detecting fraudulent transactions, predicting if an online customer will make a purchase, determining the best schedule for running advertisements and identifying potential system failures.

ScoreData, a Palo Alto, Calif.-based solution provider, recently began using H2O's platform to develop application models for predicting customer behavior in specific vertical industries, such as applications for banks to predict which customers are in danger of defaulting on their loans or applications that service companies use to identify customers most likely to churn.

ScoreData CEO and Co-Founder Vasudev Bhandarkar said in an interview that a major pain point in working with predictive models is that they need to be constantly updated. H2O's machine learning capabilities, he said, make it possible for "self-learning predictive models" that train themselves to be more accurate.

"I think H2O offers a whole, new layer of machine learning capabilities right out of the box," he said.

H2O's Ambati said his company recently recruited ScoreData and another implementation partner and plans on having as many as 20 by the end of this year.

The 3.0 release of the H2O platform offers a new web user interface with embeddable workflows, new application programming interfaces (APIs), direct integration for the Python programming language and direct integration with Sparkling Water -- the latter H2O's integration of its platform with the Apache Spark processing engine.

H2O offers an open-source community edition of its software and a commercial enterprise edition.