Predictive IT: Data Science Looks to the Future
I had the good fortune to contribute to a panel discussion at EMC World 2014 last month, focused on Predictive IT. Moderated by Bill Schmarzo, part of EMC’s Big Data consulting practice, I was joined by three other panelists intimately involved with Big Data across EMC:
- Krishnakumar Narayanan (“KK”), EMC IT’s Chief Architect and head of IT’s Data-Science-As-a-Service team
- Frank Coleman, who directs an analytics and reporting team within EMC Customer Service
- Matt Povey, who helps customers implement Big Data technology solutions
The positive feedback on the session was driven by having an engaged audience asking good questions coupled with a diverse mix of perspectives among the panelists.
Several undercurrents ran through the questions we received from the audience. First, confusion remains regarding Business Intelligence and Data Science; when to use each and when to bring these skills on to your team. This is not surprising since it’s still early days and customers need assistance in transitioning from a traditional analytics and reporting team to embark on more complex data science problem sets. Second, people also need help identifying typical IT use cases for Big Data.
Many of the questions focused on simple ways to get started with Big Data and Data Science. I shared a number of ideas on this, though I tried to boil it down to two main suggestions for those trying to move toward doing more Data Science:
- Stop thinking about the past. Start thinking about the future. Typically, reporting and Business Intelligence focus on telling you what happened last year or last quarter. Reframing questions to be about the future—What will we do next year? How are things likely to change next time, and how will we know?—cause people to think in different terms, and analyze data and situations using different methods. This shift will also require getting people on the team with the skills to apply advanced analytical methods that answer these questions.
- Adopt a “test and learn” mindset. Too often, people take the “wait and see” approach (“Let’s see how many new customers we get ….”). Instead, consider doing A/B testing or find ways to test ideas and learn from the
experiments. This critical shift toward data science needs to happen. Don’t be afraid to experiment or admit you don’t know something; that’s generally how people get out of their comfort zones and learn new things.
This Birds-of-a-Feather topic generated some thoughtful dialogue, aided by enthusiastic participation from the audience. I encourage you to visit the InFocus author pages of my fellow panelist, Frank Coleman, and our moderator, Bill Schmarzo, who regularly address interesting and timely Data Science topics on their respective blogs.