How I (almost) replaced two machine learning models with an if-statement(overencoded.com)
overencoded.com
How I (almost) replaced two machine learning models with an if-statement
https://overencoded.com/blog/how-i-almost-replaced-two-machine-learning-models-with-if-statment
3 comments
Perhaps it could have been offered as a "naive" interim solution to be replaced by machine learning, with the benefit that the ML team would have a benchmark to show how good their model was by comparison.
You'd have made the PM look like an utter fool, so he shut you down. Classic office politics. What's good for (i) the company, (ii) you, (iii) your boss... often are not the same things... sometimes in direct contradiction.
The interests of the involved parties must always be taken into account.
The interests of the involved parties must always be taken into account.
Sunk cost fallacy: they invested time, effort and their reputation
on the "smart solution", its going against the consensus.
There is no point optimizing overengineered stuff,
it will be simplified to that closed-form equation on its own,
and the company will gain some technical know-how managing
that system to implement advanced ML analytics.