Steve Mills opened up the discussion talking about Big Data, making the point that the art of the possible when it comes to data has been growing steadily for many years – though the current explosion in data is pretty impressive. For instance 1.3B RFID tags in 2005 and 30B in 2010, 4.6B mobile phones etc. The challenge is not just that we expect 40x as much data in the coming decade as we have today but that we have to find value in this data – we have to make better decisions using this data if we can. And this data is changing (more video, audio, image data) and the speed at which it arrives and changes is climbing. All this makes up the “big data” challenge for which IBM has been developing a Big Data platform based on open source foundation components, workload management and productivity tools InfoSphere Streams and InfoSphere BigInsights.
Steve gave some “big data” examples like Catalina Marketing who use 2.5petabytes of data to drive better cross-sell offers for retail consumers and use Netezza to drive model scoring time down or like BBVA who analyze 5.8terabytes of social media data to listen to their customers online or like the Telco that analyze 10terabytes of data each day to see what’s happening in their network right now and drive root cause analysis. At the end of the day, he says, support for Big Data should be part of your enterprise data platform so it can be integrated, governed as necessary etc.
Doug Hunt followed to introduce a customer talking about the use of content management in education. Randy Sumrall, the CIO of District 10 in Texas, came up to give quick summary of the need for personalized learning that helps at-risk students while driving better results overall. These objectives must be balanced against increasing class sizes and other challenges while transforming education to match the way young people thing about computers, learning and information. To make this work districts must transform their assessment, design and delivery of education. You need to create better assessment tools that allow the targeting of educational material and approach that are predicted to work for this student and then continuously assess how well this is going. To do this District 10 built an integrated warehouse/dashboard environment to inform teachers and are moving to integrate content management for teaching material, predictive analytics to show what’s likely to work and case management to manage the constant feedback so important to better educational results.
Frank Kern came up next to talk about the recent study with MIT Sloan. Clients, he says, are more and more clear that they MUST transform their businesses to survive but they must do this in a more complex, more regulated, more uncertain environment. More data, increased volatility rising expectations and continuing economic problems just pile on the challenges. The use of analytics is becoming critical – economic value no longer comes from knowing something others do not but from deriving actionable insight from large amounts of data available to everyone. The organizations in the MIT study that say analytics creates a competitive advantage are 2x more likely to be outperformers relative to their peers.
These organizations differentiate themselves in three ways using analytics. Speed of decisions, managing enterprise risk and focusing on customers.
Frank’s first example was of McKesson, using a Decision Management System with embedded analytics to make sure they could respond quickly enough for high volume, operational decisions. They had to have a system because the decision is so high volume that a 99.9% accuracy meant a loss of $100M!
Managing risk is a big issue for companies too but assessing and allowing for risk must be part of every decision, not just the big strategic decisions but operational/transactional decisions too.
With customers the move in analytic companies is to micro decisions, focusing on a single customer and making a decision just for them. Micro segmentation, personalization, incredibly fine-grained decision making.
Frank was followed by Manoj Saxena, who heads up the Watson business unit. He talked about how Watson works, how it extracts meaning from unstructured text and what this means for a change in computing from managing structured data and being deterministic to managing unstructured data while being probabilistic. Interestingly Manoj presents Watson as a Decision Support system – helping others make better decisions.
Healthcare is one of their key focus areas and is an area where information volumes are exploding – doubling every 5 years – while physicians have almost no time to read journals. Meanwhile only 20% of diagnoses are evidence-based, 1.5M prescription errors are made every year in the US and 50-100,000 people are thought to die every year due to medical errors. This seems therefore like a great opportunity for Watson-like technology. Manoj walked through how unstructured data about patients combined with unstructured medical journal data could drive Watson to help with diagnostics. Contact centers, government and financial services are also under development.
Next up was the story of using analytics to help with neo natal intensive care – Dr Carolyn McGregor came up to discuss it (I wrote about this in a white paper for IBM some time back – check out Transforming Healthcare Delivery with Analytics). Her insight was that all the data being produced by the sophisticated equipment in the NICU was being printed out and lost where it could be collected electronically and analyzed continuously. Applying analytics to the data as it streams in to flag changes in the health of a baby allows, for instance, to detect heart problems 12-24 hours before anything can be seen in the summary data. And this mindset could help with other intensive care environment as well as remote monitoring of patients (like that done by old friend Anders Bjorlin).