Two years ago, Cloudera’s Mike Olson predicted that 2013 would be the year of Big Data applications. The months have come and gone and an application ecosystem has yet to emerge, but Tresata CEO Abhi Mehta says that’s about to change. He believes the industry is finally moving up the stack.

Starting with a little background, Mehta tells theCUBE hosts John Furrier and Dave Vellante that bringing compute closer to the data – rather than the other way around – is the “biggest and most liberating concept of Big Data.” This paradigm shift is a stepping stone to the next evolution of analytics: apps.

Mehta explains that the race to zero among Hadoop distributors and other vendors at the lower layers of Big Data stack can’t go on forever. To maintain its momentum, the industry will have to migrate upward toward “advanced analytics applications,” specifically predictive analytics in Hadoop. He designates this technology as the “trend to watch” over the next four years.

Changing topics, Mehta highlights that Hadoop is meant to be a computational platform as opposed to a storage framework. He explains that Big Data pioneers such as Google and Facebook “have built successful economic models and companies around running computations at scale” using the software. “Running computations at scale requires three key parameters,” he elaborates. “You need shared architecture, you need the ability to store, process and analyze all kinds of data and then, take analytics processes and run them against all the data at the same time.” Hadoop provides these capabilities.

Where does Tresata fit in the Big Data ecosystem of tomorrow? The startup develops solutions that enable organizations to detect fraud and forecast business risk and sales trends. Mehta boasts that TL2C, a publicly traded provider of consumer scores, leverages his firm’s software to process more than 228 million records every day.

Asked to share his take on data science, Mehta says that “you cannot be a good data scientist without having business knowledge to use this massive amount of data, and make sure [that] if I brought together social, mobile, geo and transactional data what problem should I solve. You can’t answer that question without domain knowledge.”

For this reason, Mehta believes that technology can fill the knowledge gap on the other end of the spectrum. He predicts that machine learning solutions will eventually automate the entire process of aggregating and processing data, enabling business analysts to make good decisions faster.

Watch the interview below for more exclusive analysis, including Mehta’s insights into the role of the Hadoop ecosystem and the importance of data integration.