INFORMS NEWS: Roundtable focuses on big data

By Arnie Greenland

The 2013 INFORMS Roundtable Fall Meeting was held Oct. 5-6 at the Hilton Minneapolis in conjunction with the 2013 INFORMS Annual Meeting. The theme of the meeting was “big data.”

The first evening event started with a reception and dinner, which encircled the first technical presentation delivered by Claudia Perlich, chief scientist of dstillery, a New York City-based marketing services firm. Claudia, who describes herself as a “data scientist,” is a three-time winner of the KDD Cup, a data-science competition sponsored by the Association of Computing Machinery, and is on the editorial board of the new journal Big Data.

Claudia described the challenge of helping clients decide whether to bid for space on websites, based on a massive amount of data using analytics models. Claudia is a particularly engaging speaker, and she found herself inundated with questions and interest from attendees.

The second day of the event included five talks and a panel discussion. Juergen Klenk, a principal of Booz, Allen, Hamilton, and a member-representative on the Roundtable, presented the first talk. Juergen’s presentation laid out the foundations needed to understand the massiveness of the data challenge in general, but then in much more detail how it applies in the healthcare area. He discussed data challenges of electronic health records, the growing challenges of dealing with claims data such as will be generated by the new Affordable Care Act, and the emerging world of patient-generated data.

The next speaker was Rishi K. Narang, the founding principal of T2AM, an investment company that is a leader in quantitative trading. He is a veteran in the field of quantitative trading and the author of the book “Inside the Black Box: The Simple Truth About Quantitative Trading.” Rishi weaved a fascinating tale of how the truly massive amounts of trading data can be leveraged to the benefit of their clients. The talk provided another concrete example of what is meant by a “big data” problem, and it illustrated how analytics are critical to success in these problems.

The third presentation of the day was presented jointly by Mark Hayward and Chuck Martel. Mark is the administrator for the Center for Science of Healthcare and the vice chair for the Department of Facilities and Support Services of the Mayo Clinic. Chuck is the vice president for Data Management at Optum Labs. They discussed the strategic relationship between Mayo Clinic and Optum Labs. Mark began by describing the history and research and data focused culture of Mayo Clinic that dates back many decades. He also described the massiveness of the data challenges in their environment. Chuck continued the discussion by providing more details into why Optum labs was created – as a method to bring together researchers in healthcare with those in other disciplines, such as data scientists, to take on the incredible problems in this space. Chuck ended by discussing specific examples this collaboration has uncovered that are important results for clinical delivery and medical research.

After lunch, Brian Garrett of SAS presented an experiential exercise using the critical equipment of packets of M&Ms to illustrate the functioning of a HADOOP implementation. Each participate in the room became their own “HADOOP cluster” to which he “mapped” and “reduced” the results to a central server. The fun provided by the M&Ms was buttressed by Brian’s deep understanding of the tools, and all in the room came out with the clear understanding of how they work.

Dr. Dakshi Agawal, a researcher at the IBM Watson Research Labs in Yorktown Heights, N.Y., presented the last presentation of the day that was focused on yet another one of the key components of big data – streaming data. Many organizations have data streaming in from multiple (possibly hundreds or thousands) of sensors, and they need to understand and potentially act upon this data very quickly. The historical paradigm was to put this data into a database, possibly “clean it up” and then process it, but for many applications, this process would not allow action to take place in time. Dakshi described approaches and examples, primarily for the telecommunications industry, where these tools are having a great impact. As with each of the presentations, this one provided a concrete answer to the question: “What is big data all about, and how can I use it in my organization?”

The final technical event was a panel that included all of the speakers who were able to stay through the second day of the meeting. It was a very lively discussion as the Roundtable member representatives peppered the panel with a variety or questions, honing in on the understanding of how these methods could be used in their own worlds.

The meeting ended with two business activities: a short business meeting led by INFORMS Roundtable President Bill Browning and the annual Roundtable Meeting with the INFORMS Board, led by INFORMS President Anne Robinson and Roundtable President Bill Browning.