The digital business platforms of the major public cloud providers and the converged/hyper-converged hardware manufacturers have reached maturity. But, writes Wikibon Big Data & Analytics Analyst George Gilbert, they lack support for the machine learning analysis applications that play a central role in digital business.

Machine learning is vital for pulling value from the huge volumes of data associated with the Internet of Things in particular. The amount of data generated by fully instrumented machines, such as automobiles, is overwhelming without automated analysis (see image above). Such analytics applications use data feedback loops to continually refine predictive models based on data from customers, products and operations. To operate effectively they need a digital business platform that:

• Supports the movement of complex, high-volume, heterogeneous data;
• Leverages data feedback loops as machine learning’s “flywheel”; and
• Supports radical simplification of machine learning.

To do that, the platform must support:

• Application program interfaces (APIs);
• Data feedback loops;
• Operations as the cost of goods sold (COGS);
• Rapid build and test; and
• Automated deployment.

Current machine learning application layers that run on top of the digital business platforms are immature, says Gilbert. The versions that do exist are too complex for many enterprises to run effectively. The platform providers need to simplify that machine learning layer and make it an integral part of their digital business platforms.

Wikibon Premium subscribers can read the full Alert here. Learn more about subscribing here.

Image courtesy Hortonworks, Inc.