Prescriptive Analytics

A Decision Science Lens:

What is it really?

Prescriptive analytics may be the one thing in the analytic buzz universe with more hype than predictive analytics. As with everything else in this zany industry, it is over-hyped, hardly new, and heavily conflated. The reality is prescriptive analytics may only be a “thing” because organizations are quite bad at analytics — period.

How can you say that?

This series of articles uses the lens of decision science to examine other concepts in the Data Science & Analytics field. When you consider prescriptive analytics from the point of view of a decision-maker, things can have a very different perspective.

From an academic standpoint, prescriptive analytics may actually seem novel. In academic circles, analytics often stops with an insight or, at best, a prediction. The idea that analytics would also be actionable is… well, unusual.

Decision science has its roots in the military. It is a discipline and highly actionable one at that. This is also what is intended in the business world. You might expect that would be pursued with similar zeal, after all — there is money to be made. But in fact, many companies do not invest in decision science. To them, prescriptive analytics also seems novel. In reality, it is simply what should always have been.


So are Prescriptive Analytics and Decision Science identical?

That depends. Does spelling count?

In many circles, it certainly feels that way. The largest difference being the gravity of prescriptive analytics. Prescriptive analytics, like many buzzwords before it, has become so trendy that it has collected other concepts to it. Gravity. Or is that gravitas?

The central components of our cover art contain an infographic taken from Wikipedia. The graphic implies that prescriptive analytics is composed of eight other disciplines. Well actually, the graphic is not so clear. It is the caption on Wikipedia that makes the claim. Sorry, this just isn’t as clear cut as they would have you believe.

Are Predictive & Prescriptive Analytics identical?

Again no. Although they may have identical gravity. Both concepts have sucked in a myriad of other buzzwords. If you conflate two distinct concepts long enough, they become identical.

Note — Gartner’s graphic does well to distinguish predictive from prescriptive, but in turn conflates prescriptive analytics with analytics in general.

Analytics in general is the study of How. Gartner’s article does a great job with descriptive and predictive but a terrible job (IMHO) with diagnostic and prescriptive. I don’t want to get off a tangent, but any question starting with why is best left to a philosopher. Diagnostic analytics is the study of how did it happen, leave intent and philosophy out of it. Prescriptive analytics is the study of what actions should be taken to support, increase, or adjust the happening?

So if decision science is separate from prescriptive analytics, how do they work together?

Now that is the right question! Prescriptive analytics follows decision science. Once analytic discipline has been applied to determining the right priorities, the right decision framework, and the desired outcomes/direction — prescriptive analytics is used to develop a set of actions to continue the process.

Going back to our cover art, prescriptive analytics utilizes tools such as NLP and image processing to analyze and understand real-world feedback. It combines with concepts like Operations Research, an early form of decision science, to define the outcomes and directions that it will steer us toward. It is essentially a reactive concept.

So — having used decision science to determine the outcome we are seeking, prescriptive analytics defines the activities we use to get there.

Does that mean decision science sits above prescriptive analytics in the hierarchy?

Sure. It also sits below. I mean this all assumes you have any idea what you are doing. Analytics in general is highly iterative and recursive. If you think it is supposed to flow in just one direction, you are doing it wrong.

But yes, in a highly idealized world, you would likely engage in decision science first and prescriptive analytics next. But in an idealized world, nearly all analytics would be performed with outcomes and actions in mind. In that case, prescriptive analytics would feel almost redundant or be evident for what it actually is a subdivision of analytics used to direct business activities toward a known goal or outcome. It is actionable intelligence, a term that has now been relegated to the BI or reporting component of prescriptive analytics. It has been pulled in by gravity.

If it is redundant, is it meaningless?

It is a little bit of both. Add in the conflation and gravity and you really begin to wonder why anyone names anything other than to fight over it. This article has attempted to refine it to its core. If we turn off the conflation and gravity, prescriptive analytics is meaningful in its singular focus on taking action. Decision science is then focused on making a decisions and our old friend predictive analytics on forecasting likely outcomes. Along the way, they each utilize a variety of tools with interesting names. These tools are readily re-branded and have their own gravity — just to keep things confusing.

Are you going to rant about AI Overlords again?

No. Artificial Intelligence has a clear role in prescriptive analytics. Assuming the decision science has been performed, much of prescriptive analytics involves monitoring and reacting to changes. This requires the speed, processing power, and rigor that computers (and other automated systems) excel at. Prescriptive analytics is the natural home for machine learning and artificial intelligence. Eventually, this will also include Deep Learning… but again this article is getting long.

For now, understand that analytics is a very divided field, not in function but in perspective. Combine with that its complex and recursive nature and you can see how things can get very confusing. Finally, pile on the need for sales guys to make quotas, marketers to feel important, and a general human tendency to conflate almost everything. Hopefully, you have come to appreciate the need for articles like this that explain major concepts in a very direct (stylistic?) fashion.

So… thanks for reading!

read original article at——artificial_intelligence-5