Prescriptive Analytics

It’s kind of amazing how powerful our brains are, isn’t it? We can take multiple disparate pieces of information and pretty much instantly cobble them together into a cohesive opinion, plan, project, whatever. To demonstrate, let’s step inside Brad’s brain for a sec as he prepares for his afternoon commute home. Thanks to his oh-so-powerful brain, he can take five separate and distinct facts—his car needs gas, it’s been raining for two days straight, the kids’ karate lessons ended at 5:00, the oven is on the fritz, the freeway exit near his house is closed for construction—and almost effortlessly integrate them into his planned commute: he’s going to take Broadway instead of Main Street so he can hit the gas station on the way to the freeway. Then he’s going to get off that freeway two exits early so he can avoid construction traffic, stay away from that one road that always floods in heavy rains, and swing by Mickey D’s and grab some dinner for the family. Impressive, isn’t it?

Now imagine if computers could do the same thing. Wait, uh...we don’t have to imagine, because they can do the same thing. And when they do, it’s called “prescriptive analytics.”

Using A.I. and algorithms, some computers are now able to take what they’ve observed (descriptive analytics) and what they think will happen in the future (predictive analytics) and turn it into an actual course of action (this is the prescriptive part). Imagine the possibilities here: a software program that automatically order everything a restaurant needs to prepare its menu and adjusts what it orders based on price and seasonal availability. Traffic light computers that can sense and respond to changes in traffic patterns and weather and can take into consideration the fact that the next block over is temporarily closed due to a car accident.

This all sounds amazing, to be sure, but just like any computer program, the age-old maxim stands: GIGO. That stands for “garbage in, garbage out,” and it basically means that the quality of our output—in this case, our prescriptive analytics—is only going to be as good as the quality of the data and parameters we put in. As technology advances, so too do computers’ prescriptive analytics capabilities. But there are some who say that, no matter how much data a computer can crunch and how fast it can crunch it, it’ll still never be more powerful or agile than the big brains on Brad.