 |
I recently participated in a Predictive IT panel discussion at EMC World. Our panel was moderated by Bill Schmarzo (@schmarzo), part of EMC’s Big Data consulting practice. The other panelists were:
I thought this was a good mix of people with very different perspectives around this topic: Bill jumped right in, explaining to the audience that formulating questions is really the eye opening part of moving from BI to Data Science. He asked the panel their advice on what skills and training are needed and how to best approach a Data Science role. Matt was the first to answer, explaining that BI in an IT organization is all about capacity management. He said gathering all the data needed is a time-consuming, frustrating task and the data set only helps answer one question. He contrasted that with his experience with data scientists on a recent IT project. He said the data scientists worked with the operations people, opening their eyes to new ways to work with the data and pull additional value from it. His advice for preparing for a data scientist role: I completely agree with Matt’s point, especially since I’m going through a similar process right now. But I feel strongly that there’s a middle step between a BI person and a data scientist: a consultant. Once the consultant showed the path and proved the concept we could hire a data scientist…and bring in potential data scientists to be trained. Again, you bridge the gap between BI and data scientist with a consultant. Kirshnakumar “KK” Narayanan noted that although BIs are not data scientists, they have the basic capabilities to become one. KK said it’s important for BIs to get the training, especially in data visualization, to become data scientists because there is so much power and business insight driven by data science. He also identified a partnership critical to success: That led us to differentiating the roles on a data science team since I think that BI people too often thought of as data scientists.
But KK and I both agree that thinking an all-BI team is going to become data scientists is unrealistic and will lead to struggles. David Dietrich had some interesting insights into how to differentiate BIs from data scientists what kinds of qualities a data scientist should have. He also thinks getting BI people to change their thinking is to change the time horizon of how they think. David goes on to identify five main areas of data scientist capabilities.
He wrapped up by reinforcing one of my points: training was created for people aspiring to become data scientists but we also need to focus on training business leaders to have a data science understanding and perspective so they know the questions to ask to get their staff to try new things. Be sure to check out David’s blog from our panel discussion with 2 main suggestions for those looking to move toward Data Science. Just to summarize, many people in BI roles today can make the move to Data Science, as Data Science is a team sport. As I’ve said before, the Data Scientist is the hardest role to fill and requires additional training. |
