Human intervention needs to be integrated with data analytics. As I’ve written before, the technical skills of a data scientist are not enough.
So it’s interesting to see how the data analytics article in the latest issue of Harvard Business Review contrasts with what that same publication featured about Big Data just three years earlier.
The data analytics article in the latest issue of Harvard Business Review is “Algorithms Need Managers Too: Know how to get the most out of your predictive tools” by Michael Luca, Jon Kleinberg and Sendhil Mullainathan. It appears in the January 2016 issue of Harvard Business Review, where the cover has a teaser describing the article as “The Care and Feeding of Algorithms by Michael Luca et al”. The article discusses the importance of integrating human input with data analytics.
Back three years ago, Harvard Business Review did point out that Big Data still needs human insight (“Big Data the Management Revolution”, October 2012). But, back then Harvard Business Review put much greater emphasis upon the technical skills of a data scientist, though it did also acknowledge the value of good communications skills (“Data Scientist, the Sexiest Job of the 21st Century”, Harvard Business Review, October 2012). In contrast, the most recent article essentially puts its entire focus upon the importance of the human side of data and on the limitations of algorithms.
As I see it, this reflects a broader shift taking place in what is being said about data analytics. We’re seeing more and more acknowledgement of the importance of human input. People are recognizing the value of the kind of problem definition that was needed long before anyone ever heard the term Big Data. Since I am someone whose corporate job many years ago was much like what would be called a data scientist today, and I am someone who saw the limitations of algorithms back then and shifted to gaining a much broader understanding of business dynamics, I think it’s great that Harvard Business Review is now featuring the softer side of data analytics and is explaining the limitations of algorithms.
In my opinion, a key point in the article is the importance of knowing what the algorithm can and can’t tell you. As the article warns, algorithms might not be transferable to a new problem. This is important because it’s easy to think an algorithm that seems to work well in one place might always apply, even to situations that aren’t quite the same as those where the algorithm fared successfully. Human input can be vital in thinking through these situations.
Making sure the data is meaningful is so important. Yet, in my experience, it’s not unusual to talk to people who seem really excited about all the possibilities in newly available data, but who haven’t even considered that they need to think through when and why the data might not mean what you think it means.
So, in conclusion, it’s wonderful that Harvard Business Review is featuring the softer side of algorithms and pointing out limitations that algorithms have. These are important factors in getting real value from the data.
