LAS VEGAS—The boardroom is paying attention to big data analytics. And at its annual Information on Demand conference here, IBM executives were on a campaign to translate the computer science that empowers machine learning and predictive analytics into terms that C-suite executives can use in their offices.
IBM introduced a series of enhancements to its line of technology products and affiliated services tailored for specific functions like marketing and industries such as oil, retail, utilities, health care and telecommunications. The company also explained how it is building out support for the popular Hadoop open-source distributed file system that enables enterprises to incorporate unstructured data into their analytics systems.
One speaker after another at the Mandalay Bay conference center cited trends in both the economics of computing that make it possible to build analytics that scour expanding sources of changing data, and the growing hunger enterprises demonstrate to harvest that data for competitive advantage.
Big data analytics represents a third generation of computing, said Robert LeBlanc, senior vice president of middleware software for the IBM Software Group. The first enabled the processing of millions of transactions by banks and airlines. The second, sparked by the growth of the Internet and the commercial Web, saw the development of service-oriented architectures that changed the way technologists and business decision-makers thought about the utility of applications. Now we have entered the era of analytics, in which organizations are asking “how do I gather all that data up?” he said.
The answer, according to IBM, is to combine five aspects of analytics—business intelligence reporting, data exploration and visualizations, operational analytics, content analytics, predictive analytics—with three kinds of computing styles or core platforms: traditional data warehousing, stream computing to enable real-time analytics, and Hadoop.
These elements form IBM’s big data platform and the company’s products that fit into it—including InfoSphere Streams software, InfoSphere Data Explorer tools, BigInsights software for social media analytics, among others—received updates. Some updates build on previously released offerings, some use technologies from acquired companies (Data Explorer builds on Vivisimo technology), while others are designed to optimize Hadoop for corporate computing environments.
A specific offering aimed at chief marketing officers, Digital Analytics Accelerator, is designed to analyze consumers’ sentiment from social media, smartphones and other mobile devices and traditional channels. The company also announced a new technology and service offering for health care providers.
In addition, IBM unveiled a cloud offering called Analytics Answers aimed at small- and medium-sized businesses to access predictive analytics via the Smart Cloud subscription service. Earlier this month, IBM announced new hardware appliances called PureData Systems, tuned for functions like transactions, operational analytics and in-memory analytics using Netezza technology.
Other tech giants, including Oracle, SAP and Teradata have developed their own big data platforms and methods for helping organizations looking to build out their analytics capabilities. IBM, though, has publicly stated it is investing about $16 billion in its R&D programs and acquisitions.
One company that says it has benefited from those investments is Conoco Philips, the large oil and gas producer. Phil Anno, principal scientist, told about 12,000 attendees that his team of data scientists uses InfoSphere Streams as an application development environment up on which to build predictive models for ice floes in the Arctic Sea, a frontier in oil exploration.
The Conoco Philips models use satellite images to track several hundred ice floes, generating about one terabyte of data per day. Researchers believe that their forecasts will yield an extra month in the summer drilling season, Anno said. More testing and trials are required, but the real-time data streaming environment took a chunk of systems work off the hands of Conoco Philips employees’ time, meaning they can spend time developing other data-driven experiments to design oil rigs that can work in the Arctic elements.
Discussions of other use cases played a prominent role at the conference, as IBM executives delivered tutorials on the business value of analytics.
For example, Martin Wildberger, vice president of information management development, outlined four themes for enterprises to tackle: “Know everything about your customer” by analyzing data to manage customer churn, social media sentiment, and the next-best action for call center operators to take. Target operational efficiency, through predictive analytics applied to maintenance, supply chain optimization, and network management. Mitigate risk and fraud, through fraud modeling and detection and risk modeling. And create new business models, through data analytics experiments, such as location-based advertising.
All of this opens up new opportunities for service providers analyzing their data streams, Wildberger said. The question to ask, he said, is “How can I provide new services going forward?”