We're behind the scenes again for Predictive Analytics World! Tapan Patel and I recently caught up with James Taylor, CEO, Decision Management Solutions for a quick Q&A session on Predictive Analytics World that takes place during Data Driven Busines Week March 14-15 in San Francisco.
Any thoughts you would like to share for upcoming Predictive Analytics World 2011 Conference? What role are you playing at the conference?
James Taylor: I am doing a couple of things at the conference. First, the day before the conference kicks off, I am running a workshop on Driving Enterprise Decisions with Business Analytics. This is an all-day session on the whys and hows of decision management and using it to apply predictive analytics to operational systems. Monday the event kicks off and I will be staffing my company booth – Decision Management Solutions is exhibiting for the first time at Predictive Analytics World and I am hoping to get a chance to talk to attendees about their use of analytics in operations, especially at the evening reception.
Tuesday morning I will be listening to Tom Davenport’s keynote and then participating in the International Institute of Analytics (IIA) Executive Forum. I am a faculty member of the IIA and this is our first event in conjunction with Predictive Analytics World. I will be moderating a roundtable on Critical Issues in Applying Analytics at Production Scale. Finally I wrap up the show with my presentation on Deploying analytics with a rules based infrastructure. I will be using some of my clients as cases to illustrate why and how a business rules management system, and an integrated approach to business rules and analytics, can be effective in deploying predictive analytics.
What are the top reasons for attendees to sign up for Driving Enterprise Decisions with Business Analytics workshop?
Taylor: Well obviously the first and most important is that they get to spend the whole day with me! More seriously, the workshop is focused on how to approach the problems of applying advanced, predictive analytics in operational systems. Far too many analytic models, especially those that relate to day-to-day operations, fail to get deployed and used effectively. The workshop lays out the reasons for this and how an explicit focus on operational decisions – the micro decisions that affect one customer or one transaction – results in more effective operations with analytics. Attendees come to learn what operational decisions are and how to find them, to learn a little bit more about some of the technologies that can be used like business rules, and to get an overview of an effective methodology for addressing the challenges of applying analytics in automated systems.
What do you think are some of the common challenges customers face before embarking on a predictive analytics project? How will the full-day workshop address some of the challenges?
Taylor: I see several common challenges. The first is that predictive models just make predictions and must be tied to action to add value. I see too many organizations begin by trying to see what they can predict instead of beginning by determining what they could do to “move the dial” of their business. The workshop will show how beginning with the decision in mind can address this and tie predictive analytics to specific decisions so that it is clear what actions can and should be taken to turn those predictions into valuable action.
The second is that applying analytics in production means getting the business, IT and analytic teams to work together – something they do poorly if at all. The workshop will discuss the reasons for this and some of the technologies and approaches that can mitigate it.
Thirdly I see a lack of focus on what I call “industrializing” analytics. Organizations of any size need an industrial-scale analytic approach if they are to really change their business to one driven “by the numbers”. The workshop is really all about this kind of process and providing a methodology for doing analytics in a more systematic way.
Finally companies often think that analytic projects are one-offs where the evidence is that it is ongoing, continuous improvement that adds the most value. With a strong focus on Decision Analysis and on adopting test-and-learn approaches, the workshop encourages attendees to build in continuous improvement right from the start.
Tell us a little bit about the Moderated Roundtable Discussion you’ll be leading for the IIA Executive Forum, “Critical Issues in Applying Analytics at Production Scale.”
Taylor: This is new event for IIA and should be really interesting. A limited number of people can participate in each roundtable as we want to keep these sessions really interactive. I will kick it off by talking about some of the challenges I see in my clients when they are working on applying predictive analytics in operational, high volume-low latency environments. Then I will moderate a discussion, encouraging those present to talk about their issues and potentially how they resolved them. The idea is for the group to bring up, discuss and find solutions to the problems that the group brings to the discussion. Highly interactive and focused on real problems it should give those who come concrete advise on how to address their issues. I am really looking forward to it.
Based on your customer engagements, what are some of the hot new areas in predictive analytics offering the most potential in next 3 to 5 years?
Taylor: I think there are many areas that are interesting. The idea that different kinds of analytics (structured, unstructured, web etc) are all separate is going to be replaced with a decision-centric approach of “what do I want to know so I can act more effectively” that will pull all these different kinds of analytics together. I think the integration of text analytics with structured data analytics is going to be hot particularly.
Ensemble models, and the ability to generate all the component models very quickly, is another area with potential. The evidence for the success of ensemble models is real so tooling to take data and rapidly produce lots of “basic” models that can be combined in an ensemble is needed and will be popular.
I also think companies are going to be more and more focused on how to get more models, updated more often with fewer staff. Tools to bring non-statisticians into the mix and help make it easier to rapidly build and update models will be a big deal as companies industrialize their analytic processes and push analytics out to every decision in every business process.
Special thanks to James and Tapan for taking time to talk to me about Predictive Analytics World. Please stay tuned for more pre-event interviews regarding the upcoming March 14-15 conference! Be sure to let me know what you want to hear more about prior to Predictive Analytics World or send me something you'd like to share on sascom voices.
