A Trip into Data Analytics

Based on the Introduction to Data Analytics course taken on Coursera.

Learning to think about analytical problems
The best analysts are people who really love understanding how things work and can create clarity out of chaos.
Knowledge and experience are the keys to the analytical process, but something that happens often is that the analysis doesn’t lead anymore, meaning that it doesn’t stimulate action or change an action that would have taken place anyway. There is a key distinction between the analytics we do in the business world and the analysis that might be done in science in academia. In business, we need a clear return on the investment we make in analytical people and the tools that support them. Knowledge alone doesn’t justify an investment unless we can show it makes a difference in the decisions we make to run the business.

A business analyst operates in a resource-constrained environment. There is never enough time or money to do all the things we’d like to do. We need to be focused on determining what things we do and how we do them.

There is a simple principle that we can apply to make sure that our analytical process is efficient and effective—specifically, we think backward, meaning we start with the decision we want to make, then we consider the information needed to make that decision. It is closely related to the hypothesis-driven approach used in scientific research, where we start with something specific we think is true, then we design an experiment to test the hypothesis that it really is true.

Ways to go about it

Context is needed; a person needs to understand enough about the business and about how decisions are made to understand what’s really important and what’s most likely to sway a decision one way or the other.

The first question to ask yourself: Can I perform an analysis that will actually influence a meaningful decision? If you can say something like, if my analysis shows outcome X, I’ll do one thing, and if it shows outcome Y, I’ll do something else, and the choice I make matters, then you’re in good shape. If you discover that you will probably make the same decision regardless of the outcome or that the decision itself really isn’t all that important, then you should reconsider investing in an analysis.

The second question to think about is what the output of our analysis would look like. What story would I expect to be able to tell? How would I actually see Outcome X or Outcome Y in the data? Is there a specific chart or graph that would illustrate the difference?
The more specific you can describe the output, the better. You might even sketch out what your final presentation will look like in advance.

With a clear view of the output in mind, you can start thinking about the analysis itself. What methods do you need to apply? What tools will you need?
There are different ways to work with data; some of them are simple, others complex. A good guiding principle is that simpler is often better. What is the most straightforward way of getting to the answer? We should know when to use complex methods at times.

To be continued……