SAS interview: How advanced analytics is changing the worldBy Ryan Noik 30 January 2018 | Categories: interviews
In a fascinating interview with Paul Kent, the vice President of Big Data initiatives at SAS, the industry veteran explained the potentials that advanced analytics holds when applied to data and just how influential the cloud is to business, and people’s day to day lives
It is no secret that data has become exponentially more important as we have progressed in the 21st century. But, as the focus has shifted to gleaning value from data, the role of analytics has become more significant.
To begin with, Kent explained the difference between traditional analytics and advanced analytics, explaining that the former is what is achieved using spreadsheets, summaries, statistics, rollouts, by region.
Advanced analytics, however, is different. On the extreme end, he elaborated, an example of advanced analytics is the underlying mathematics that run in the cameras of autonomous cars that will drive themselves in the future. It will be advanced analytics that would be responsible for collecting data such as the video feed, radar, and the analyses of images, so as to differentiate between a stop sign and a tree, and determine whether a car may be about to swerve into one’s lane.
But advanced analytics isn’t confined to the developments to come. On a simpler level, it is already being utilised to enable businesses to cluster their customer base, and market differently to each cluster. “So we think of advanced analytics as basic mathematics, statistics, forecasting and optimisation,” Kent summed up.
Not a static state
Even so, Kent pointed out that analytics are always evolving. “The computing power that is needed to perform sophisticated image recognition simply was not practical fifteen years ago, but CPUs and GPUs have evolved. These advances in technology enable us to do more than we could previously,” he enthused.
For example, the recommendations engine on Xbox for example, which suggest other types of media based on a game you are currently playing, is actually a massive computational problem. He pointed out that the engine has to serve millions of users and analyse hundreds of thousands different artefacts - either games, or song, or videos - while matching the preferences of one player with a variety of others to compare games that they had played. Such analyses could easily require 10 000 computers to “crunch” through the data.
“In the past, firstly companies would never have had that much detail about the problem, and secondly, they wouldn't have had that much computing power to handle it. But now they do. It just happens to be the same computers that do the storage and the compute and so approaches that used to be considered extreme are now attainable,” he added.
Accelerated pace of change
Along with the improved capabilities now at our disposal, Kent pointed out that change and innovation have become considerably more rapid, rather than slow and steady, due in no small part to the cloud. He elaborated that in fact the cloud encourages continuous innovation, since developers can roll out a myriad of minor improvements whenever they desire.
“It is rumoured that Facebook makes thousands of small changes a day. Many of them are AB tests, whereby they send 50% of their users to one version of the change and 50% to a different version of it and then they choose which they are going to continue with. They are continuously doing that and gradually everything gets better,” he noted.
How to save a life
Admittedly, not every organisation is the size of Facebook or has reams of engineers at their disposal. However, that doesn’t mean that, along with the cloud, advanced analytics cannot be used to create relevant and potentially life-saving solutions.
Kent cites the example of one unnamed developer who is building an image recognition model that uses a medical database of images that identifies which skin lesions are cancerous, could become cancerous or are merely cosmetic. While the analytics involved are highly computing intensive, the model itself can be distributed as an app and downloadable to a smartphone. Users would then be able to point their device’s camera at a freckle or mole, for example, and the app would be able to match it against the database and let them know whether they can safely ignore it or, or whether they need to see a specialist.
“That's a doable one-man project that is in the works, and the cloud has facilitated its realisation in two ways. Firstly, it enabled the developer to take advantage of a provision of computing power to build the model to make it work. Secondly, it has enabled certain parameters to be run on the user’s device in order to determine whether their skin lesion was dangerous or not,” he elaborated.
So which industries can we expect to see advanced analytics really make headway? One where big data and advanced analytics are being eagerly embraced, according to Kent, is in the industrial arena. He elaborates that in airplane engines, for example, sensors are being used to ‘tell’ Boeing whether a particular part needs to be repaired before it breaks. The cost of doing maintenance, he points out, is considerably less than dealing with actual breakage, that spews metal and causes further damage, not to mention risking the lives of passengers.
The health industry is just getting used to what it can do with sensors, data and the analysis of that data, to provide better healthcare. “These applications are in the early days, they are still working on what is acceptable and what is not, and how much smarter can you make the sensor. I think there will be plenty of developments as these kinds of ideas mature,” he adds.
Even as these examples are exciting and illustrate the relevance advanced analytics (along with big data and the cloud) has to our emerging world, there are impedances that need to be overcome in order for advanced analytics to be used to its fullest extent.
Kent points out that the lack of skills is and education is a problem. “There are only so many top shelf engineers and they are scooped up by Google, Facebook and Uber, so getting the skills distributed across everybody is a challenge across many technologies,” he concludes.
Most Read Articles
Have Your Say
What was the top tech news story of 2017?