Now that Summertime is upon us, the time is there, to consider what you have been up to. And what your New Business Year’s Resolutions are for when people … sorry I mean co-workers, return from their well-deserved holidays away from you. So that next round of business activity, you’ll do better.
We need to have a talk.
I mean, qua data processing the new way, as in ML. In the global data swamp, how come good data is hard to come by? Why do we still struggle to get the right data, in sufficient quantities?
[Let alone the biggo problem of integrating the developed system’let into the landscape of old monolithic operational systems, that actually do something.]
We blogged earlier about WEC. Which assumed one has the right data, in the right quantities; only the quality per data point needed tweaking.
But then, what happens when one doesn’t have quite the right data, but a couple of proxies here and there… What will you be doing …?? And what will you do with the outcomes …?
Like, this: Your time series will (sic) have massive distortions unless you watch your step very, very carefully. And this, even worse.
But worst of all, … being in a tunnel qua vision, and not seeing the wider context. Then, biases and other total system distortions like too-weak-proxy errors will occur. And not only render your efforts much less useful if at all, but also will discredit your work/methodology and that of all others in the field as well.
So … Are you feeding your quants the right info? Qua context and (relevance of) data? Qua relevance of the system altogether [beyond mere fitness in the IT landscape]?
Reconsider … In the end, They‘ll find the one common factor among all fail factors re your AI project(s), being You.