In 2012, a framework was published in a leading journal that talked about predictive modelling.
It said we need to be problem solvers.
"Products rang[ing] from weather forecasting to recommendation engines to services that predict airline flight times [can do so] more accurately than the airline itself. But these products are still just making predictions, rather than asking what action they want someone to take as a result of a prediction."
This is such an interesting time in machine learning and algorithm-ing, because of that very opportunity - the ability to exceed human capability in a narrowly-defined valuable context. But it is limited where we talk about it in the minutiae alone.
The great example is the Amazon recommendation algorithm that, as a best-in-class solution, is often/actually/still not any good at predicting the things we want to buy next, because it suggests things we already have or are likely to already be aware of.
"Solving the problem" feels like a different challenge than "improving the solution", and now, in the data economy especially, we are hitting a golden age for "potential problem solving." An algorithm is a "solution" (perhaps one for identifying my early-stage glaucoma). An algorithm that improves upon my doctors' ability to recognize the glaucoma from a scan is an "improved solution" than the one we've got (and that's exciting).
But a data-driven technology that takes me out of the medical system flow and still cures me, or prevents my ever contracting the malady? That's Solving the problem. Where can data solve entire problems?
Yours in HAT,