My Biggest Gripe With Perscriptive Analytics

Witten by Melissa Castellanos (May 21, 2025)

Let’s be real, you don’t need a data science degree to figure out what predictive modeling means. It’s right there in the name; you’re creating a model to predict possible outcomes. According to/ the school of Google, predictive modeling is a statistical technique used to forecast future outcomes by analyzing patterns in existing data. In other words, it’s using math to guess what’ll happen next.

It seems fancy and like the hobby of some finance bro in a Patagonia vest, but predictive modeling is everywhere. From the 401(K)s most Gen Z will never have, to the never-ending list of Amazon’s “You Might Also Like” suggestions, we’re all interacting with these statistical models more than we realize.

So, what’s the big deal? A few more bucks spent on your Amazon cart doesn’t hurt anyone, except maybe your bank account, right? WRONG. Predictive models aren’t just fueling shopping sprees; they’re being used to make decisions that actually matter, like in banks, governments, and private companies. And these decisions have serious consequences.

Despite, and maybe because, predictive models are basically a set of mathematical equations, results aren’t always accurate. When you create a predictive model you have to quantify uncertainties, and how uncertain is an uncertainty if you can quantify it? In other words, if you are already thinking about something causing issues, these issues are no longer uncertain because you are expecting them to some extent. What happens to the issues you don’t expect?

Accurate or not, predictive models are paraded around like some kind of futuristic oracle without mention of how flimsy they actually are. My biggest issue? They create a false sense certainty for anyone who isn’t actively questioning them. It seems like the more we rely on these models, the more we seem to crave a glimpse of the future. It’s like we’re addicted to prediction, even when the foundation is shaky.