Keith Collins, Sr. Vice President and Chief Technology Officer of SAS, opened his plenary talk at INFORMS today reporting the dream he woke up to this morning. In that sleepy state upon awakening he pictured researchers on Antarctica, and as consciousness came to him he tried to determine the signficance. Was it about the mountains of data from core samples, sensors, temperatures, etc. or simply about being a computer scientist, a lonely island tasked with speaking to a sea of operations researchers? His talk turned out to be about both, since he addressed the phenomenon of the deluge of data from the perspective of a computer scientist who leads an R&D team focused on finding new ways to glean insights from this deluge.

So what trends does Keith see from his Land of Computer Science? Whether you call it Big Data or not (he doesn’t like the term, since what will the next generation call it – Bigger Data?), analytics is more important than ever. Our existing approaches to dealing with this data won’t work, so he encouraged collaboration across disciplines to integrate techniques for more robust solutions. And he said three things he’s especially excited about are the possibilities for new insights from high performance computing, visualization, and text analytics.

He gave a shout-out to INFORMS for broadening its reach from a traditional operations research audience and embracing the entirety of advanced analytics. The experience SAS has had facing the problems customers bring to us has called us to make this same move. One example was markdown optimization for a major retailer. They needed to know which products to discount when by SKU at a store level, which involves combining a complex estimation challenge with optimization. Their immense scale of 3 terabytes of data used to take 30 hours to run, but with in-memory distributed processing and high performance analytics the window came down to 2 hours, allowing models to be rerun with several iterations by Monday morning instead of one time per weekend.

Next Keith addressed visualization, which can be most useful to test a hypothesis and prevent business analysts from having to hunt and peck for an answer. He gave an example of social network analysis applied to preventing fraud at the Los Angeles County Department of Social Services. Taxpayers saved $31 million in the child welfare public assistance program, aided by a visual network that could quickly and easily cue investigators toward uncovering fraud rings.

The final trend Keith spoke about was changes in the kinds of data we are seeing and the value of text analytics to understand it. The number of transactions/person is no longer increasing, but the attributes/transaction is, which means growing amounts of unstructured data. Warranty analysis looking at product complaints, health outbreaks predicted looking at EMS logs, and consumer buzz about products can all be understood far more effectively with text analysis.

Keith closed by speaking to the importance of relationships between industry and academia. As interest in analytics grows the demand for talent does as well, so programs are needed to turn out ever more talent with the necessary skills. And he charged the audience with helping reform computer science education by helping them understand the value of getting information out of data in addition to teaching how we get the data in.

NOTE: This blog post was originally published on the Informs Transformation blog.