New frontiers for analytics
SAS CTO Keith Collins on the hottest new applications for analytics - and what's coming next
Our customers around the world tell me that there is more data than ever, leading to greater expectations for decisions and less confidence in making those decisions based on intuition alone. Analytics has never been more important.
David Mohler, CTO of Duke Energy, says the network is key to the smart grid. As major home devices become IP-enabled, the network will be flooded with data that contains huge potential for both customers and society. Interesting questions arise about how to define layers of optimization from the home, to the transformer, to the distribution grid, including determining the minimum data necessary to automate appropriate actions at each stage (check out the sascom article smart grid).
Eric Bibelnieks, Group Manager of Guest Data and Analytical Services at Target, shared his belief that high-performance analytics brings tremendous opportunities to optimize across problems that were historically rendered independent, suboptimal activities because of their complexity and scale.
Eric Williams, CIO at Catalina Marketing, manages what might be the world's largest single database table to tailor unique offerings for any American who uses a grocery loyalty card. The volume of data is amazing, but even more astounding are the revenue results he gets for his customers by leveraging advanced analytics.
These conversations illustrate the growing importance of analytics. Solving problems requires practitioners to move beyond their core discipline to call upon a full range of analytical practices. We look forward to seeing new areas of applications across industries and across analytics.
Where it all began
Over time SAS has moved from selling software tools for statistics, OR, data mining and econometrics to more directly developing solutions for our customers' business problems. We still make and sell these tools, now stronger than ever, but we are also combining underlying algorithms in particular ways to find answers that were previously considered unattainable. One of our motivations is that the power of analytical techniques is now accessible to users who might not be experts in the specific techniques. There simply are not enough experts, and the opportunities are so compelling that a broader audience wants to try analytics.
A good example is our work in operations research. We first released SAS/OR® software in 1982, and in recent years we have completely revamped our offerings in optimization and simulation. Drawn by a broader market understanding of the power of OR, we have been applying underlying OR algorithms to specific problem classes, such as marketing optimization, and even in specific industries, such as revenue management and price optimization in the hospitality industry. This evolution has involved gaining industry expertise to understand questions such as what kind of data is available, what is and isn't easy to measure, and what regulatory and business process constraints apply. All these questions vary by industry.
This evolution also means combining different analytical approaches. The tight margins in retail make maximizing product profitability at the end of a season or the end of product life critical. Profit maximization requires knowing which products to mark down when, by how much, and in which locations. Attaining this information requires estimating demand as a function of price; this is not an easy estimation problem because available data is typically not from demand, but from sales. Because the problem is so closely tied to the optimization, the best solution is an integrated approach, such as the one offered in SAS Markdown Optimization for the retail industry.
As we bring our approaches closer to the business and more accessible to a broader audience, we are also maintaining the analytical rigor through R&D investments. We realize we can’t develop software for every problem, so we staffed an OR Center of Excellence with individuals who have both doctoral-level training and also years of experience applying OR techniques to problems across industries. They focus on model formulation, model solution and implementation. They have helped customers solve problems in industries such as manufacturing, retail, consumer packaged goods, banking, media and entertainment. They draw on their own experience as well as deep expertise across the Advanced Analytics Division. Due to positive customer response, we are currently exploring additional Centers of Excellence dedicated to industries, such as manufacturing, and to applications,
I see three other exciting frontiers: high-performance analytics, visualization and text analytics.
–Keith Collins, SAS CTO
Surging interest in analytics requires new ways to convey these insights to laypeople. Business visualization displays analytics in compelling ways to a wider range of data consumers than ever before, because a picture can truly be worth a thousand words. Users across an organization can have a single, consolidated view and can also focus on views based on their roles. Within a business intelligence framework, these insights can be centrally and securely administered and then delivered to end users via the Web or even mobile devices. And using JMP® software from SAS, the visualization can be interactive and exploratory, so you can quickly get to the most important insights your data has to offer.
Most experts estimate that 70 percent of data is in unstructured, nonnumeric format, so SAS has stepped up investment in our tools for text mining, content categorization and sentiment analysis. We have also combined techniques to solve specific business problems; for example, our new SAS Social Media Analytics enables organizations to gain detailed insight into chatter across various social media sites and then attribute these threads to specific parts of their business.
Our employees keep us abreast of these emerging trends so we can maintain our own competitive edge. In addition to participating in ongoing training and study, they attend conferences and are active in various professional organizations. For example, Kathy Lange is President-Elect of the INFORMS Roundtable and Bob Rodriguez will be President-Elect of the American Statistical Association (ASA) in 2011. In addition to these roles, many other employees are in leadership roles or on committees of INFORMS, ASA and other professional organizations. We challenge our employees to translate new approaches they encounter into innovative ways to solve emerging problems. Because I’m quite confident that the problems are only going to get harder. And SAS will be here with solutions.
A version of this article originally appeared in the Sept/Oct 2010 issue of Analytics magazine, a publication of INFORMS, the Institute for Operations Research and the Management Sciences.
Bio: Keith Collins is a Senior Vice President and Chief Technology Officer for SAS.
This story appears in the First Quarter 2011 issue of