Citizen Analytics — It All Starts With A Question

Whether you are a casual observer or a seasoned professional in data analytics the sheer number of different types of data analytics talked about these days will certainly make your head spin. Here are just a few of the main types you routinely see mentioned all the time by analysts, journalists and resident data scientists (where each one has several naming variations):

  • Diagnostic Analytics
  • Descriptive Analytics
  • Prescriptive Analytics
  • Predictive Analytics
  • Exploratory Analytics
  • Visual Analytics

And so on… At this point most of us know it when we see it — a marketing “gone wild”:

Descriptive vs. Predictive vs. Prescriptive

These three are typically hailed as the main categories for data analytics in general. Descriptive Analytics is looking back and telling you what has happened already, Predictive Analytics supposed to tell you what will happen in the future, and Prescriptive Analytics should tell you what to do about that specifically in your business.

However, if you look closely, all three differ only in a way they interpret the analysis of the existing data, in other words — how they use the result of the actual data analysis. Since, obviously, we can’t analyze the future data, all three approaches are based on analyzing existing data to answer some question that underlined one of these types of analysis.

Once this answer is obtained (i.e. we figured out what has happened in the past), we can statistically extrapolate it into the future (predictive usage) and, potentially, map this future data extrapolation to the business terms (prescriptive usage). Leaving aside the accuracy and overall value of the prescriptive use — both predictive and prescriptive analytics are nothing more than an additional step after the main data analysis is done.

In a essence — the descriptive, predictive and prescriptive data analytics are fundamentally the same process with a variation of how the end result is interpreted and presented.

We can only analyze the data we have. After that we can draw different conclusion from it — what has happened, what may happen and how we may need to react.

It All Starts With A Question

At DataLingvo we believe that any data analysis starts with a… question. Simple or complex, frequent or ad-hoc, spontaneous or well planned, important or just a hunch to test — a question about the data is what triggers the data analysis.

Think about every time you open Google Analytics dashboard, Salesforce.com reporting pane, Tableau IDE or fire up SAS. Even if you just do you daily monitoring of your AdWords campaign — you are answering a simple question: “Did anything significantly change in my AdWords campaigns since yesterday?” You don’t directly ask that question, but the AdWords dashboard you are staring at is just a convenient way to answer that frequently (daily) asked question.

Ability to easily ask a question about any of your business’ data and get an answer that you can effectively interpret is at the core of our ability to derive business value from all the business data we are collecting today.

I believe we are over complicating the meaning of data analytics for no apparent benefits to the end users. This over complication creates artificial shelving and product categorizations just for the sake of shelving and categorization itself.

It all starts with a question. Data analytics is a system that gets you answer to that question. That’s as simple as that.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.