Big Data: Trade Offs Necessary
September 14, 2015
i read “How to Balance the Five Analytic Dimensions.” The article presents information which reminded me of a college professor’s introductory lecture about data analysis.
The basics are definitely important. As the economy slips toward 2016, the notion of trade offs is an important one to keep in mind. According to the article, making sense of data via analytics involves understanding and balancing:
- The complexity of the data. Yep, data are often complex.
- Speed. Yep, getting results when the outputs are needed is important.
- The complexity of the analytics. Yep, adding a column of numbers and calculating the mean may be easy but not what the data doctor ordered.
- Accuracy and precision. The idea is that some outputs may be inappropriate for the task at hand. In theory, results should be accurate, or at least accurate enough.
- Data size. Yep, crunching lots of data can entail a number of “big” and “complex” tasks.
I agree with these points.
The problem is that the success of a big or small data project with simple or complex analytics is different from a laundry list of points to keep in mind. Knowing the five points is helpful if one is taking a test in a junior college information management class.
The write up does not address the rock upon which many analytics project crashes; that is:
What are the systems and methods for balancing resources across these five dimensions?
Without addressing this fundamental question, how can good decisions be made when the foundation is assumed to be level and stable?
Most analytics work just like the textbook said they would. The outputs are often baloney because the underlying assumptions were assumed to be spot on.
Why not just guess and skip the lecture? I know. Is this an acceptable answer: “That’s too time consuming and above our pay grade”?
The professional who offers this answer may get an A in class but an F in decision making.
Stephen E Arnold, September 14, 2015